The Swiss linguist Ferdinand de Saussure (1857-1913) has ex-

erted a profound influence not only on twentieth-century lin-

guistics but also on a whole range of disciplines in the humanities
and social sciences. His central thesis was that the primary object
in studying a language is the state of that language at a particular
time - a so-called synchronic study. He went on to claim that a
language state is a socially constituted system of signs which are
quite arbitrary and which can be defined only in terms of their
relationship within the system. This new perspective has changed
the way people think about linguistics and has led to important
attempts to apply structuralist ideas in anthropology, literary
criticism, and philosophy.
Professor Holdcroft's book expounds and elaborates Saus-
sure's central ideas. He also offers a critical assessment of them,
arguing that many of Saussure's claims are either questionable
or have been misunderstood. The book will be read with profit
by non-specialists and could be used as a textbook by students
of linguistics, philosophy of language, literary criticism, and
anthropology.
 SAUSSURE
SIGNS, SYSTEM, AND ARBITRARINESS
 MODERN EUROPEAN PHILOSOPHY
Executive editor
RAYMOND GEUSS, COLUMBIA UNIVERSITY

Editorial board
HIDE I S H I G U R O , KEIO U N I V E R S I T Y , J A P A N
ALAN MONTEFIORE, BALLIOL COLLEGE, OXFORD
MARY TILES, UNIVERSITY OF HAWAII

R. M. Chisholm, Brentano and Intrinsic Value

Maud Marie Clark, Nietzsche on Truth and Philosophy
Raymond Geuss, The Idea of a Critical Theory:
Habermas and the Frankfurt School
Gary Gutting, Michel Foucaulfs Archaeology of Scientific Reason
Karel Lambert, Meinong and the Principle of Independence
Frederich Neuhouser, Fichte's Theory of Subjectivity
Charles Taylor, Hegel and Modern Society
Mary Tiles, Bachelard: Science and Objectivity
Robert S. Tragesser, Husserl and Realism in Logic and
Mathematics
Peter Winch, Simone Weil: The Just Balance
 SAUSSURE
SIGNS, SYSTEM, AND ARBITRARINESS

DAVID HOLDCROFT
Professor of Philosophy, University of Leeds

The right of the

University of Cambridge
to print and sell
all manner of books
was granted by
Henry VIII in 1534.
The University has printed
and published continuously
since 1584.

CAMBRIDGE UNIVERSITY PRESS

CAMBRIDGE
NEW YORK PORT CHESTER MELBOURNE SYDNEY
 CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, Sao Paulo

Cambridge University Press

The Edinburgh Building, Cambridge CB2 2RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org
Information on this title: www.cambridge.org/9780521326186

© Cambridge University Press 1991

This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.

First published 1991

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication data

Holdcroft, David.
Saussure: signs, system, and arbitrariness / David Holdcroft.
p. cm. (Modern European philosophy)
Includes bibliographical references (p. ) and index.
ISBN 0-521-32618-4 (he). ISBN 0-521-33918-9
1. Saussure, Ferdinand de, 1857-1913. 2. Linguistics. I. Title. II. Series.
P85.S18H6 1991
410'.92-dc20 90-48917
CIP

ISBN-13 978-0-521-32618-6 hardback

ISBN-10 0-521-32618-4 hardback

ISBN-13 978-0-521-33918-6 paperback

ISBN-10 0-521-33918-9 paperback

Transferred to digital printing 2006

Extracts and drawings from the Baskin translation of CLG are

reprinted with the kind permission of Peter Owen Ltd.:
Publishers.
 For

EILEEN MARY
 CONTENTS

Preface page ix
Introduction 1
1 Saussure's work: its context and significance 4
2 The distinction between langue and parole 19
3 Language as a system of signs, I: Signs,
arbitrariness, linearity, and change 47
4 Language as a system of signs, II: Diachronic and
synchronic linguistics 69
5 Language as a system of signs, III: Identities,
system, and relations 88
6 Language as a system of signs, IV: Values,
differences, and reality 107
7 Successes and failures 134
Notes 161
B ibliography 175
Index 178

Vll
 PREFACE

I owe my interest in Saussure to Willie Haas, Professor of Lin-

guistics at the University of Manchester from 1963 to 1979.
Attendance at his lectures stimulated me to read the Course in
General Linguistics, and I discussed it and structural linguistics in
general with him on many enjoyable and profitable occasions.
Even though I was not convinced by the structuralist method-
ology to which he was deeply wedded, I learnt a great deal from
him; above all I remember his patience with, and forbearance
of, views which attracted him not at all. Later, at the University
of Warwick, I renewed that interest whilst giving a course on
semiology and structuralism, which led me to appreciate Saus-
sure's achievements not only as a linguist but also as someone
whose ideas had had a major impact on the human sciences in
general. The origin of the material in this book dates from lec-
tures I gave at that time, and I owe a debt to students who
attended those lectures at Warwick, Leeds, and Stony Brook for
their contributions to what turned out to be an unexpectedly
enjoyable course from my point of view.
I owe a debt too to colleagues at Leeds both for letting me off
with a light teaching load in the autumn of 1987, which enabled
me to start writing this book, and for a number of helpful dis-
cussions in seminars since then. These discussions, together with
ix
x PREFACE
ones at the universities of Aberdeen, Edinburgh, and York
helped me to clarify a number of issues. I would also like to
thank the Department of Philosophy at the State University of
New York at Stony Brook for providing a stimulating and re-
laxed atmosphere in which to work when I visited in the spring
of 1988; drafts of several of the early chapters date from that
time. Finally, my thanks to the editors of this series for their
encouragement and help.

Leeds
 INTRODUCTION

The Course in General Linguistics is not an easy book to read.

Apart from the fact that for much of the time Saussure was
preoccupied with fundamental issues about the nature of lan-
guage and the methodology of linguistics, the text itself is not
Saussure's but one reconstructed by Charles Bally and Albert
Sechehaye from students' notes of the three lecture courses in
general linguistics that he gave. The difficulties of reconstruction
they faced were formidable (1.3); and though their achievement
was an extraordinary one, the structure of the resulting text
does not make it easy to follow Saussure's argument. This apart,
he himself clearly did not think his theory as it stood was ready
for publication. If at times his argument seems incomplete or
inconclusive to the reader, so too it did to its author, though one
can, for the most part, only guess in what respects it seemed
deficient to him. Furthermore, the text has been a victim of its
own success. Saussure's claim that the primary linguistic study
is the study of a state of a language as it is constituted at a
particular time (a so-called synchronic study) rather than one of
its evolution through time (a 'diachronic' study) has come to be
so widely accepted that it is hard to see how anyone could have
thought otherwise. Yet much of CLG has to be seen as a critique
of nineteenth-century historical and comparative linguistics,
2 SAUSSURE
which did indeed assume that one can engage in historical and
comparative studies without grounding them in synchronic ones
(1.1). This negative thrust of Saussure's argument is puzzling to
a modern reader unaware of the context in which he wrote.
Accordingly, I have tried, in Chapter 1, to provide a brief
description of the context in which CLG was written and of some
of the major concerns that motivated Saussure. This account not
only provides a background for the study of the text, but also
outlines some of the central issues to be tackled, and tries to
show what was novel and exciting about Saussure's treatment of
them. Of course many of these issues recur frequently through-
out this book, making it difficult to keep track of Saussure's
overall position, so in Chapter 7 we return for a final assessment
of these central themes in the light of the intervening discussion.
Chapter 1 concludes with a brief discussion of the text of CLG
itself and of the very special difficulties of interpretation it poses.
The editors claim that the order of the text is substantially that
of the last course of lectures Saussure gave on general linguistics,
the so-called Third Course; but the reality is rather more com-
plicated, and they in fact departed from his order in a number
of ways. In my view, the actual order of the Third Course is for
the most part substantially clearer than the one they adopted,
and the exposition of Saussure's argument in Chapters 2 to 6 of
this book follows that order for this reason.
The five chapters devoted to the exposition of Saussure's ar-
gument trace the development of his central theses: first, that
the primary object of linguistic study is a synchronic one, that
is, one of the state of a language at a given time; and second,
that a given language state is a system of arbitrary signs whose
signifying properties depend entirely on their place within the
system. Thus, three of the pivotal notions of that argument are
those of a sign, of a system, and of arbitrariness. That is why I
have used the title 'Language as a System of Signs' as the overall
chapter heading for Chapters 3 to 6. Moreover - apart from the
fact that if we take it as read that the signs are arbitrary, many
of Saussure's major themes are neatly encapsulated - the title is
one which was proposed by Saussure himself for the chapter of
CLG in which he embarks on his theory of the sign (Part 1,
Chapter 1), though it was not in fact used by the editors.
The overall argument is undoubtedly complex, and the reader
may think that its exposition has not been helped by the presence
 INTRODUCTION 3
of critical comments as it unfolds. Could they not have been
saved to the end? Perhaps they could; but not very easily, since
the point of many of the criticisms is precisely to clarify just what
Saussure has to prove at a given stage to move his argument
along a further step. Anyway, each chapter, except the last, con-
cludes with a summary; and strung together, the summaries
form a kind of abstract of the full discussion which will, I hope,
make it easy both to relate a detailed discussion to a wider per-
spective and to see how it relates to the development of the
overall argument. Further, to make it as easy as possible to trace
the discussion of a particular theme, I have frequently cross-
referenced the current one to earlier and later ones, even though
the reader will no doubt discover that I do not always say the
same thing in different places. Thus, the \i.$Y in the first par-
agraph of this introduction refers to Chapter 1, Section 3.
 SAUSSURE'S WORK
Its context and significance

That the work of Saussure, a Swiss linguist, should be the subject

of a book in a series called Modern European Thought should
occasion little surprise, for his subject was one to which he made
a seminal contribution. What is notable is that he is an important
figure not only in the development of twentieth-century lin-
guistics, but also in that of European philosophy and the human
sciences. What then distinguishes Saussure from other linguists
of his generation whose work had no such impact?
Perhaps the pre-eminent difference is his concern for foun-
dational issues, unclarity about which had been, he thought, the
source of much confusion in the practice of comparative and
historical linguists - the dominant nineteenth-century tradition
- so much so that further progress depended on their clarifi-
cation. Connected, he believed that the new conception of lan-
guage that he proposed was important not simply because it
cleared up confusions which had hindered the progress of lin-
guistics itself, but also because it avoided mistakes which it was
all too easy to make in the philosophy of language. So he believed
that his work had important implications not only for linguistics
but for philosophy too.
At first his philosophical morals had very little impact outside
 SAUSSURE S WORK 5
linguistics itself. But, as Mounin says, Saussure's work, however
belatedly, affected 'and unquestionably enriched the thought of
such thinkers as Merleau-Ponty, Levi-Strauss, Henri Lefebvre,
Roland Barthes, Lacan, Michel Foucault, and through them all
the contemporary social sciences' (1968, 9). It is true that Mou-
nin's claim about the impact of Saussure's work on the social
sciences has to be understood to apply primarily to the continent
of Europe; but its importance there for the development of
structuralist thought and methodologies is unquestionable.
Often the sort of enrichment of later thought that Mounin speaks
of occurs without the original thinker having any prevision of
the ways in which his thought might be developed in other areas
by other thinkers. But this was not so in Saussure's case. He saw
what he took to be the possibility of widening the scope of his
methods by treating linguistics as merely a species of something
more general, namely a semiological system.
For Saussure, a semiological system is any 'system of signs that
express ideas' (CLG, 33, 16).J Natural languages are examples
of such systems, but so are 'the alphabet of deaf-mutes, symbolic
rites, polite formulas, military signals, etc' Since a language is
only one kind of such a system, even if it is the most important
kind,

a science that studies the life of signs within society is conceivable; it

would be a part of social psychology and consequently of general
psychology; I shall call it semiology (from the Greek semeion 'sign').
... Linguistics is only a part of the general science of semiology;
the laws discovered by semiology will be applicable to linguistics,
and the latter will circumscribe a well-defined area within the mass
of anthropological facts. (Ibid.)

This claim was, of course, programmatic. The laws in question

were still to be discovered, so that the newly envisaged science
did not then exist. But there is no doubt that in trying to define
the place of linguistics in the human sciences (fig. 1.1) Saussure
posited a wider space in which new enquiries might in turn find
their place. However, exactly what the boundaries of this space
were and what enquiries could be located there are an almost
entirely conjectural matter in CLG; and, as we shall see later,
Saussure's embryonic conception of semiology contained a num-
ber of unresolved tensions (7.4).
 SAUSSURE

General psychology

Individual psychology

[Etc.

Linguistics Military signals Writing

Figure 1.1

Undoubtedly, Saussure's main claim to our interest is his con-

cern with fundamental questions about the nature of language
and the methodology of linguistics, and above all with the ques-
tion of the nature of the object of linguistics itself. To understand
why these questions seemed to be of such importance to him it
is necessary to say something about the methodology of nine-
teenth-century linguistics, the tradition he grew up in and ulti-
mately found wanting.

1.1. Historical concerns

In the nineteenth century, linguistics had been dominated by
comparative studies to which Saussure himself made a notable
contribution, including the work which established his reputa-
tion, A Memoir on the Primitive Vowel Systems in Indo-European
Languages', and it is worth remembering that he remained keenly
interested in comparative and historical questions throughout
his career. The first chapter of CLG is devoted to a very brief
survey of the history of linguistics, which, says Saussure, 'passed
through three stages before finding its true and unique object'
(CLG, 13, 1). Comparative linguistics itself corresponds to the
third of these stages; moreover, his discussion of the previous
stages is so brief by comparison that it is reasonable to infer that
he thought that the work of the comparativists was the first
empirical work done by linguists which merited serious consid-
eration.2
 SAUSSURE S WORK 7
Commenting on Bopp, the founder of this tradition, Saussure
writes:

While Bopp cannot be credited with the discovery that Sanscrit

is related to certain languages of Europe and Asia, he did realize
that the comparison of related languages could become the subject
matter of an independent science. To illuminate one language by
means of another, to explain the forms of one through the forms
of the other, that is what no one had done before him. (CLG,

But although Saussure respected the aims of comparative lin-

guists, believing that they had identified a legitimate and im-
portant field of study, he came to think that their work was
methodologically confused:

The first mistake of the comparative philologists was also the

source of all their other mistakes. In their investigations (which
embraced only the Indo-European languages), they never asked
themselves the meaning of their comparisons or the significance
of the relations that they discovered. (CLG, 16, 3)

Clarity demanded, Saussure came to think, that a linguist be

clear whether he is dealing with different forms of the same
linguistic item, e.g., the same word pronounced differently, or
with different items, however closely they might resemble each
other, e.g., a pair of homonyms (2.1.1). Comparisons made,
often over considerable periods of time, which neglected to ad-
dress these fundamental questions could well go completely
astray. This in fact happened, Saussure believed, in the work of
Schleicher, a comparativist who had an organicist conception of
language and hence believed that languages belonging to the
same kind - e.g., Greek and Sanscrit - all have to go through
the same developmental stages, in the same way as plants be-
longing to the same species do: ' . . . languages are natural or-
ganisms, which without being determinable by the will of man,
grew and developed in accordance with fixed laws' (Schleicher
1869, 21). Such a view was, Saussure thought, incredible. As a
result of it, language was considered a specific sphere, a fourth
natural kingdom; this led to methods of reasoning that would
have caused astonishment in other sciences' {CLG, 17, 4).
As we shall see, Saussure developed, in opposition to this view
8 SAUSSURE
of a language as a natural object evolving in accordance with
fixed laws, a conception of language as a social product, succes-
sive stages of which come about without individual or collective
design (4.2, 7.3). In CLG he repeatedly emphasizes the social
nature of languages, which he describes as a 'social fact' (CLG,
21,6) which is 'outside the individual who can never create nor
modify it by himself; it exists only by virtue of a sort of contract
signed by the members of a community' (CLG, 31, 14). Elsewhere
he says, 'Contrary to all appearances, language never exists apart
from the social fact' and 'its social nature is one of its inner
characteristics' (CLG, 112, 77).
Mounin has argued that this sociological dimension of Saus-
sure's thought is one which has to be defined in relation to
Durkheim. But, as we shall see (7.3), the relation of Saussure's
ideas to Durkheim's is by no means straightforward, and the
issue of the balance between individual and holistic elements in
Saussure's thought is a delicate one.
A further inadequacy of the comparativist's approach emerges
when one reflects on the fact that it was restricted to only one
kind of explanation, a comparative and historical one. But how
could this be ultimately satisfying? To explain the significance
of a word in terms of its derivation from a word in a different
language, or from a word in an earlier stage of the same lan-
guage, merely raises the question of how the significance of the
form from which they are derived is to be explained. If all we
are restricted to is a historical explanation, then an infinite re-
gress seems to be unavoidable.
Thus, because it failed to ask fundamental questions about its
object of study, and in particular failed to ask what was being
compared with what, comparative linguistics was at bottom meth-
odologically confused. In addition, the ontological commitments
of the organicist conception were in Saussure's view incredible.
This conception degenerated into the speculations of Schleicher,
which treat natural languages as though they are members of a
species of natural object.
In the fourth stage of the history of linguistics, with which
Saussure seems to have identified himself at least in part, some
of the grosser confusions of the comparativists had, he thought,
been cleared up. The American linguist William Dwight Whit-
ney, who had compared a language to a social institution resting
on convention, had, Saussure thought, made an important con-
 SAUSSURE S WORK 9
tribution. So too had the Neo-Grammarians, a school of German
scholars with positivist inclinations, whose contribution had been
in placing 'the results of comparative studies in their historical
perspective, and thus linking the facts in their natural order'
(CLG, 19, 5). In other words, they had not simply contented
themselves with comparing different states of languages, how-
ever remote from each other they were, but had also tried to
describe linguistic changes in detail and to propose mechanisms
to account for them. Moreover, they had rejected the organicist
conception of a language, so that 'thanks to them language is no
longer looked upon as an organism that develops independently
but as a product of the collective mind of linguistic groups' (ibid.).
As we saw, Saussure himself was a keen protagonist of the
view that language is social; so presumably what he meant when
he said that it was in this fourth and final stage that linguistics
had found its true object was that it was no longer regarded, as
Schleicher had regarded it, as a species of natural object, but
as a social product. But lest the reader should think that there
was no more fundamental work to be done, he adds that 'the
neo-grammarians did not illuminate the whole question, and the
fundamental problems of general linguistics still await solution'
(ibid.). In other words, the clouds obliterating the landscape had
lifted, but its detailed features still remained obscure, awaiting
the illumination to be provided by the subsequent chapters of
CLG, which make a decisive break with the then prevailing view
that historical linguistics is the fundamental study.

1.2. Saussure and the philosophers

At first sight, the idea that Saussure addressed himself to phil-
osophical issues is somewhat strained; a glance at the index of
CLG shows that it does not even include 'philosophy' as a head-
ing. Nevertheless it is possible to see a number of points at which
Saussure was keenly aware of the philosophical relevance of his
work. Saussure's brief history of linguistics begins by telling us
that at the first of the four stages that he distinguishes 'something
called "grammar" was studied. This study, initiated by the Greeks
and continued mainly by the French, was based on logic' {CLG,
13, 1). The study in question was regarded by contemporaries
as primarily philosophical. This was something he had no need
to point out, since his audience would of course have understood
1O SAUSSURE

the reference to French work of this sort as one to the tradition

of general grammar, which began with the publication of Lan-
celot and Arnaud's Grammaire generate et raisonnee in 1660, often
known simply as the Port Royal Grammar, and continued until
the end of the eighteenth century.3 So Saussure's brusque dis-
missal of this tradition is an index of the extent to which he
distanced himself from it. Its faults were in his view many: 'It
lacked a scientific approach and was detached from language
itself. Its only aim was to give rules for distinguishing between
correct and incorrect forms; it was a normative discipline, and
its scope was limited' {CLG, 13, 1).
It is true that later this resolute criticism is tempered by the
acknowledgment that the tradition adopted an irreproachable
viewpoint from which to study language: 'Their program was
strictly synchronic. The Port Royal Grammar, for example, at-
tempts to describe the state of French under Louis XIV and to
determine its values.... The method was then correct, but this
does not mean that its application was perfect' (CLG, 118, 82).
However, though he thought that the viewpoint of traditional
grammar was irreproachable, since its aim was to describe the
state of a language at a fixed time rather than the way in which
it had changed through time, his objections to its methodology
and assumptions were deeply rooted. As well as the objections
cited above — its normativeness, its limited scope — there was,
Saussure thought, a fundamental one. This was that the tradi-
tional grammar tried to derive the structure of sentences from
the logical form of thoughts, so that, for instance, to the logical
categories of substance and quality there correspond the gram-
matical categories of noun and adjective (Ducrot 1968, 19). But
such a view, Saussure came to believe, failed to recognise the
autonomy of language. The structure of a language is, he ar-
gued, autonomous in the sense that it is internal to itself and
not a reflection or representation of something else, e.g., the struc-
ture of thoughts or of independently given facts. Since language
is autonomous, it is a cardinal mistake to try to explain features
of its structure in terms of features of other structures. This kind
of mistake, Saussure thought, is all too prevalent, and is one
which philosophers of language are especially prone to make.
In an interesting passage at the beginning of Part 1, Chapter
1, of CLG Saussure criticises the view that language is essentially
 SAUSSURE S WORK 11
a nomenclature, i.e., the view that words are simply labels of
independently identifiable things:

This conception is open to criticism at several points. It assumes

that ready-made ideas exist before words...; it does not tell
whether a name is vocal or psychological in nature (arbor, for
instance, can be considered from either viewpoint); finally it lets
us assume that the linking of a name and a thing is a very sim-
ple operation — an assumption that is anything but true. (CLG,
97> 65)

To be sure, the points are made so briefly that it is difficult to be

certain what weight to attach to them. So it is an important point
that this is in fact a very condensed version of a note in Saussure's
own hand addressed specifically to philosophers (Engler 2,
147F).
Most thinkers about language, he argues in this note, approach
it as though it were a nomenclature. The majority of philosoph-
ical conceptions make one think of Adam calling the animals
before him and giving them names. But the conception of lan-
guage as a nomenclature has three grave defects. To begin with,

the basis of language is not constituted by names. It is an accident

when a linguistic sign is found to correspond to a perceptual
object, such as a horse, the fire, the sun, <rather than an idea such
as 'il posa'>. Whatever is the importance of this case there is no
evident reason to take it as typical of language. (Engler 2, 148F)

But not only does the nomenclaturist view of language go astray

because most words are not names and do not designate things,
it also assumes that there exists

first the object, then the sign; therefore (what we always deny),
that there is an external basis to the sign, and that language can
be represented as follows:

-a
objects * b names
* r

although the true representation is

12 SAUSSURE
a b c

apart from all <knowledge of an effective relation, such as

* a based on an object>. (Ibid.)

In other words, nomenclaturism goes astray not simply because

most signs are not names, but more fundamentally because its
account of the significance of a sign is not intralinguistic, that is
in terms of its relations to other signs, but extralinguistic, in
terms of its relation to something independently identifiable. In
giving this kind of account, nomenclaturism impugns the au-
tonomy of language. In the passage cited, the external objects
in question are perceptual ones; however, Saussure makes it clear
that he is resolutely opposed to the idea that signs have an ex-
ternal basis of any kind. So he objected just as strongly to the
kind of determination of signs by 'elements' of thoughts envis-
aged by the Port Royal Grammar.
A third mistake which Saussure thought nomenclaturism
made was that it totally ignored the effect of time. For the
changes wrought by time affect not only the words used to des-
ignate a thing or idea — e.g., the Latin calidum becomes the French
chaud (warm) — but also the very ideas they express — e.g., the
Latin crimen (accusation) becomes our 'crime'. But if nomencla-
turism were correct and words were merely labels for indepen-
dently identifiable things, then, Saussure thought, this should
not happen; only the words (= labels) should change (Amacker

As they stand, the arguments are too abbreviated to make

serious evaluation profitable; the third, for instance, seems to be
an objection to a Platonistic version of nomenclaturism which
few would espouse. But there is no doubt that the first two
arguments sketched here are ones which Saussure reverts to time
and time again, amplifying them at many points, when discussing
the principle of the Arbitrariness of the Sign and its ramifications
(3.2). The third argument is developed in great detail in his
discussion of linguistic change (Chapter 4) and indeed seems to
be involved in a separate if little-noticed argument for the ar-
bitrariness of the sign.4
Thus, not only did Saussure dissociate himself very sharply
indeed from nomenclaturism, but he also believed that the tra-
ditional grammar had not freed itself from this conception, since
 SAUSSURE S WORK 13
it tried to base itself on something 'given'. Moreover, in so doing
it impugned the autonomy of language, as did all other theories
which tried to explain features of signs in terms of the features
of independently identifiable extralinguistic entities. Evidently a
radically new approach was called for.

1.3. The writing of CLG

In their introduction to CLG the editors explain the problem
they had in preparing the text. When Saussure died in 1913
they had hoped to find extensive notes used by him for the
courses in general linguistics that he gave in 1907, 1908-9, and
1910-11 - his only lectures on this topic. But they were disap-
pointed, finding 'nothing - or almost nothing - that resembled
his students' notebooks' (CLG, 7, xxix). Moreover, they had not
attended his lectures themselves - if they had, their expecta-
tion of finding extensive notes might well have been different,
since it seems that often Saussure came armed with no more
than little pieces of paper. However, though the lectures were
attended by relatively few students, they had kept unusually
detailed notes. Four students supplied the editors with their
notes of the first two courses, and three with their notes for the
third course. These notes, together with some additional ma-
terial, provided the editors with a different problem from that
arising from the absence of Saussure's own notes, for there was
if anything almost too much material. The notes of Riedlinger,
for instance, their only source for the First Course as well as
the most detailed for the Second Course, total 732 pages. As
they say, 'to publish everything in the original form was impos-
sible' (CLG, 8, xxx).
After considering a number of alternatives, they decided to
take the Third Course as their starting point,
using all the other materials at our disposal, including the personal
notes of F. de Saussure, as supplementary sources At each
point we had to get to the crux of each particular thought by
trying to see its definitive form in the light of the whole system.
We hadfirstto weed out variations and irregularities characteristic
of oral delivery, then tofitthe thought into its natural framework
and present each part of it in the order intended by the author
even when his intention, not always apparent, had to be surmised.
(CLG, 9, xxxi)
14 SAUSSURE
However, the impression left by this, that though the text of
CLG is not Saussure's its relation to the surviving records of his
teaching is relatively straightforward, is very misleading. For
publication of Godel's Les sources manuscrites du cours de linguistique
generale in 1957 and of Engler's critical edition of CLG in 1967
makes it clear just how bold the work of the editors had been,
so much so that one commentator, Calvet, has argued that the
resulting text gives a completely misleading impression of Saus-
sure's project (1975, 54).
The editors' initiatives are grouped by Calvet under three
headings. The first of these, unsurprisingly, raises questions
about the heterogeneity of their sources and the way in
which they have amalgamated them. A good example of this
sort of difficulty would be Part 1, Chapter 1, of CLG, which
is based on the notes of four students as well as on a long
manuscript note of Saussure's. The editors' task was compli-
cated by the fact that, having had second thoughts about the
order of presentation, Saussure proposed a new title for the
chapter in question, namely 'La langue comme systeme de
signes' and also the replacement of a pair of contrasting
terms which he had originally used, 'acoustic image' and
'concept', by the terms signifiant (signifier) and signifie (signi-
fied) (3.1.1). So as well as having to decide how to incorpo-
rate four different sources into a unitary text, the editors
also had to decide both whether to adopt the new suggestion
for a chapter heading, which oddly enough they did not do,
and how to replace the terms 'acoustic image' and 'concept'
with the proposed terms. It is therefore far from surprising
that, as de Mauro points out, the resulting text is a mixture
of the two terminologies, and by no means always a happy
one5 (TM, 439).
The second type of editorial initiative involves adding ma-
terial of their own to which nothing in the sources seems to
correspond. It is perhaps surprising, but there seem to be
sentences in the text which have no known analogues in the
published sources. These include the famous closing remark:

From the incursions we have made into the borderlands of our

science, one lesson stands out. It is wholly negative, but it is the
more interesting because it agrees with the fundamental idea of
 SAUSSURE S WORK 15
this course: the true and unique object of linguistics is language (la
langue) studied in and for itself (CLG, 317, 232)

Doubtless nearly every sentence of their text presented the ed-

itors with delicate decisions. They surely had at times to make
additions to make explicit what was implicit in the sources, or
to amalgamate them into a coherent text (TM, 407). Moreover,
they were right to do so in the light of an overall interpretation
of Saussure's thought. But it was surely quite another matter to
incorporate into the text summaries of that interpretation, how-
ever justified the summary may have been. Normal editorial
practice would be to include such summaries in a preface or in
footnotes flagged 'Eds.'
Moreover, as de Mauro points out, this particular interpre-
tation was at the same time both highly tendentious and influential
(TM, 476). To be sure, one could argue that it is not without
warrant, citing the famous passage from the very outset we must
put both feet on the ground of language (la langue) and use language
as the norm of all other manifestations of speech (langage)' (CLG, 25,
9). But even if this passage were itself unproblematical, which it
is not,6 it is quite different to argue that a study of la langue
should form the basis of all studies of language, whether they
be comparativist, geographical, or something else, than it is to
argue that language (la langue) should be studied for its own
sake. On either account a study of la langue is central, but whereas
the first account does not deny the existence of other legitimate
points of view from which language can be studied, the second
does.
The third type of editorial initiative raises questions about the
overall structure of CLG. Here the real burden of complaint
against the editors is not that they changed the order of the
Third Course; it is rather that though they leave the impression
that it has been observed and that material from elsewhere has
been used only to supplement it, in fact they depart from it at
many points. Consulting SM shows that CLG makes use of ma-
terial from all three of the courses, presenting it in an order
which follows none of them very closely (SM, 103).
In fact, as Calvet points out, broadly speaking the actual order
the editors adopted inverts that of the Third Course. After a
preliminary lecture devoted to the history of linguistics and a
16 SAUSSURE
discussion of the object of linguistics, Saussure's next lecture
outlines the overall structure of his course:

November 4th. General division of the course:

1. Languages. (Les langues.)
2. Language. (La langue.)
3. The language faculty (langage) and its exercise by
individuals.
Justification of this order: the difficulty of determining the con-
crete object of linguistics. To begin with, it is necessary to separate
from the faculty of language (langage) language itself (la langue),
a social product, a semiological institution: There is the object of
linguistics. But this social product manifests itself through a great
variety of languages. It is therefore necessary to begin with what
is given: languages; then to extract what is universal: la langue.
Only then will one concern oneself with the individual's use of
language. (SM, 77)*

But whereas the Third Course begins with extensive reflections

on the plurality of languages, geographical diversity and its
causes, and so on, CLG relegates nearly all of this material to
Part 4, a part which many who have read the book to see what
Saussure has to say about synchronic linguistics will at best have
skimmed; yet, except for an interlude for a discussion of phon-
ology, this material took up Saussure's lectures from 4 November
1910 to 25 April 1911. Furthermore, even if one confines oneself
to sections dealing only with synchronic issues, there are unex-
plained departures from the Third Course.8
Further discussion of textual issues would be out of place here.
But this brief survey shows that, at the very least, great caution
is needed when citing the text of CLG as evidence of Saussure's
views.9

1.4. Summary and prospect

By training a comparative linguist, Saussure became dissatisfied
with the theoretical and methodological assumptions underlying
the practice of the field (1.1). Though he never doubted that it
was legitimate to attempt to describe the evolution and devel-
opment of particular languages, and in so doing relate them to
other languages, he believed that comparative linguists were al-
ways in danger of going seriously astray because of their failure
 SAUSSURE'S WORK 17

to ask fundamental questions, in particular what precisely is

being compared with what. This brings up a fundamental ques-
tion: What are the units to be studied and what are their identity
conditions?
Thus Saussure's dissatisfaction with historical linguistics led to
a concerted effort to answer the question 'What is the object of
linguistics?' - undoubtedly the central question addressed in
CLG. Moreover, his answer to it is a striking reversal of the
intellectual perspective in which he had grown up, since he claims
that the primary study is a descriptive one of the state of a
language at a given time, a synchronic study, rather than a study
of the changes undergone by a language through time, a dia-
chronic study. Now that his views on this matter have become the
norm, it is difficult to appreciate how both novel and revolu-
tionary his position was, yet it unquestionably represented a ma-
jor shift in perspective, however incompletely it may have been
worked out in detail.
The question of the object of linguistics was central in another
respect. For in the course of expanding his answers Saussure
developed a conception of language as a social institution in
conscious opposition to the organicist conceptions of it to which
he objected so vehemently (1.1). This conception stresses the
autonomy of language; that is, maintains that the forms of pos-
sible linguistic representations are not constrained by anything
external or prior. Clearly the Port Royal Grammar did not ac-
knowledge the autonomy of language (1.2), since though lan-
guage is, according to it, structured, its structure is a reflection of
something else: thought. Moreover, the characteristic failing of
philosophical approaches to language is nomenclaturism, the
assumption that words serve simply as labels for independently
identifiable things (1.2). But since, as Saussure saw, it is difficult
to avoid some form of nomenclaturism, the rejection of any form
of it is a radical move, calling for the development of a very
different conception of language than that customarily held.
If the conception of a language as an autonomous social in-
stitution, sui generis precisely because it is autonomous, was a
second radical idea that Saussure grappled with in CLG, there
is arguably a third, namely that of the study of language as part
of a more general study of sign systems, semiology. But though
this idea has been extremely important for the development of
structuralism, it is not obviously consistent with the idea of a
18 SAUSSURE
language as an autonomous social institution. Since semiology
as such did not exist when Saussure wrote, this tension might
not seem to matter, since it could be argued that little is lost in
the development of Saussure's system by giving up the semio-
logical approach. I shall discuss this response later in more detail
(7.4); but for the time being it may stand provisionally, since it
is clear that a considerable part of Saussure's argument does not
depend essentially on adopting a semiological approach.
The ways in which Saussure attempts to develop these themes
in detail, and the problems that arise in doing so, form the
substance of the remainder of the book. We begin in the next
chapter, as Saussure did, with a discussion of the central ques-
tion: What is the object of linguistics?
 THE DISTINCTION BETWEEN
LANGUE AND PAROLE

As Saussure himself did, I begin the systematic exposition of his

views with the distinction between langue and parole. But why
begin here? One reason is the unquestionable importance of the
distinction for Saussure's system. However, given his overall se-
miological approach (7.4), the question arises whether the dis-
tinction could in fact be derived from some fundamental
principles of semiology, in which case it would be natural to
begin with them. This is a difficult question, which will be dis-
cussed in Chapter 7, where it will be argued that there is no
convincing derivation of this kind, so that there is no alternative
to treating it as an independent distinction.1
The overall aim of Chapter 3 of the Introduction to CLG is
to answer the question 'What is both the integral and concrete
object of linguistics?' (CLG, 23, 7), so a major part of Saussure's
interest in drawing the distinction is to enable him to give a
precise answer to that question. Moreover, it is clear that the
concept of langue plays a crucial role in the answer he gives.
What is disputable is not this, but the nature of that role. Is
langue itself definitive of the object, so that linguistics is confined
to a study of it and nothing else? Or is it that without which we
cannot define an object of study, which includes langue of course,
but other things as well?2 I shall argue that the second of these
2O SAUSSURE

Langage

Figure 2.1

alternatives was the one Saussure maintained, whereas the fa-

mous final remarks of CLG, which we saw are the editors' and
not Saussure's, attribute the first view to him (1.3).
One final comment. Though the subject matter of this chapter
is a dichotomy, in fact three terms are being distinguished, name-
ly natural language — langage; a particular language — langue;
and speech —parole.5 Indeed, Saussure spends if anything more
time trying to distinguish langue from langage than he does trying
to distinguish it from parole. Unfortunately, the distinctions Saus-
sure is trying to make in French are far from easy to express in
English, so I shall continue in this section to try to express Saus-
sure's thoughts as naturally as I can, whilst at the same time
citing the corresponding French terms.4

2.1. The nature of the distinction

How then are language (langue) and speech (parole) related to
each other, and both to natural language (langage)? The central
idea seems to be that the latter can be divided exhaustively into
two sub-domains, namely language and speech. Following Godel,
this relation can be represented as in Figure 2.1 (1957, 153)- If
we think of this as a semiological definition, then langue and
parole are being explained in terms of their relations with each
other, by means of a set of contrasts:

Langue Parole

Social Individual
Essential Contingent
No active individual role Active role
Not designed Designed
 LANGUE AND PAROLE
Conventional
Furnishes a homogeneous
subject matter for a branch
of social psychology
21
Not conventional
Furnishes a heterogeneous
subject matter studied by
different disciplines

More will be said in the next section about each of these contrasts
and Saussure's reasons for making it, together with the distinc-
tion it underpins.

2.1.1. The basis of the distinction

Saussure's argument begins by considering the view that lin-
guistics should concern itself just with sounds. If someone utters
the French word nu, are we not presented with a concrete object
that can be studied?
Saussure's response to this suggestion is that whether we are
considering a sound, a word, or something else depends on what
kind of thing we have in mind: Is it a sound, the expression of
an idea, an equivalent of the Latin nudum? (CLG, 23, 8). But in
the latter two cases, in which what we are considering is indeed
a linguistic item, we have to consider it not simply as a sound
but as an instantiation of an element of a language. So to consider
an utterance of nu as one of a word is not to consider it as a
purely physical event, in the way in which one might consider
'Ah' uttered by someone as purely an involuntary exclamation.
The sounds we make when speaking, considered just as sounds,
can be described by someone who does not know to which lan-
guage they belong. To describe them as words, one must know
what features of them are functionally significant from the point
of view of the language to which they belong.5
Perhaps the best way to convince oneself of the importance
of linguistic knowledge for the identification of a word is to note
that someone who does not speak English will hear the sequence
of sounds [passthesugar] as a continuous sequence in which there
are no natural divisions, whereas an English speaker will hear it
as a three-word sentence, Tass the sugar.' Knowledge of English
enables him or her to segment the sequence into three meaning-
bearing units, which it is impossible to do without such knowl-
edge.6 It is, I think, this point that Saussure is making when he
writes:
22 SAUSSURE
But suppose that sound were a simple thing: would it constitute
speech (langage)? No, it is only an instrument of thought; by itself
it has no existence. At this point a new and redoubtable relation-
ship arises: a sound, a complex acoustical—vocal unit, combines in
turn with an idea to form a complex physiological-psychological
unit. (CLG, 24, 8)

In other words, since sound is only the instrument of thought,

the sorts of sounds linguistics is interested in are not simply
sounds, but sounds that have meaning, or functionally significant
units of such sounds.
The issue raised in this passage - what is involved in the iden-
tification of a word - is for Saussure an absolutely central ques-
tion. Failure to realise its importance and to answer it
satisfactorily lay, he believed, at the root of the confusions of
comparative linguistics, and it is a question to which he returns
repeatedly in CLG (1.1). The answer he gives at this point, that
a word cannot be identified with a sound as such but has to be
considered as a meaning-bearing item, is, of course, only a pre-
liminary answer; but the negative thesis that linguistics is not
concerned with sounds qua sounds is one that stands without
further qualification.
Before proposing his alternative to the view that language is
constituted by sounds, Saussure makes two claims about natural
language (langage) which it is essential to keep in mind. First, it
has both a 'social and an individual side'; and second, it 'always
implies both an established system and an evolution; at every
moment it is both an existing institution and a product of the
past' (CLG, 24, 8).
But if we must take all of these factors into consideration, the
search for the object of linguistics is made even harder:

... if we fix our attention on only one side of each problem, we

run the risk of failing to perceive the dualities pointed out above;
on the other hand, if we study speech (langage) from several
viewpoints simultaneously, the object of linguistics appears to us
as a confused mass of heterogeneous and unrelated things. (CLG,
24,9)

In other words, to concentrate on just one aspect, e.g., the social,

would be to ignore the others, the individual, the institutional,
 LANGUE AND PAROLE 23
and the evolutionary; whilst to study all aspects simultaneously
would leave no unifying core.
Saussure's solution to this dilemma is bold. It is to award langue
the pre-eminent place in the study of natural language (langage),
subordinating the other studiable aspects to it. It can be awarded
this place, Saussure argues, for a number of reasons: It admits
of an independent definition; it is a self-contained whole; and
it is a principle of classification (CLG, 25, 9). This last claim is
puzzling if taken to mean that langue itself provides a principle
for classifying the other aspects of natural language. But what
I suggest it means is not that, but rather that it provides a prin-
ciple for classifying linguistic units, such as nu? This would in-
deed be a reason for awarding it a pre-eminent place and for
subordinating other studies to it if, as seems plausible, they had
to presuppose the results of such classifications. An interest in
speech acts, for instance, might lead one to note that the choice
between 'you' and T as agent in the schema ' will go' can
make the difference between a directive and a commissive inter-
pretation, e.g., between a request and a promise. But such an
observation obviously assumes that we already know how to an-
alyse sentences structurally.
But what is langue} At this stage Saussure's characterisation is
tantalisingly brief:
It is not to be confused with human speech [langage], of which it
is only a definite part, though certainly an essential one. It is both
a social product of the faculty of speech and a collection of nec-
essary conventions that have been adopted by a social body to
permit individuals to exercise that faculty. (CLG, 25, 9)
His response to an imagined objection explains in part what he
means. The objection is that, after all, when we speak we make
use of a natural faculty, our vocal apparatus, whereas langue is
constituted by convention. To this he replies that whilst Whitney
may well have gone too far in arguing that we might just as well
have used gestures and visual symbols rather than acoustical
ones, he is right to insist that the actual nature of the sign is a
matter of secondary importance. Appealing to a definition of
'articulated speech' as the subdivision of a chain of meanings
into significant units, he argues that what is natural to man is
not 'oral speech but the faculty of constructing a language' (CLG,
26, 10). Presumably he thinks that, by comparison, the study of
24 SAUSSURE
the means used to articulate a language is of importance sec-
ondary to a study of the language itself. For he goes on to adduce
findings of Broca which, he argues, show that various oral dis-
orders are attributable not to the malfunctioning of specific or-
gans, but to the malfunction of a general faculty, so that they
affect writing as well:8 'The obvious implication is that beyond
the functioning of the various organs there exists a more general
faculty which governs signs and which would be the linguistic
faculty proper' (CLG, 27, 11). So what we should be interested
in is the nature of the underlying faculty rather than any one
of the particular instruments, such as speech used to articu-
late it.
But how does this faculty relate to langue} It is surely not clear
that langue can be identified with the faculty of language, since
what is common to people who speak different languages is the
possession of that faculty. But if it cannot be so identified, then
the question arises whether the object of our study should be
the latter rather than the former. I shall discuss later what Saus-
sure's answer to this objection might have been (2.3). However,
a hint of an answer is contained in his final argument for giving
a pre-eminent place to langue, namely that 'the faculty of artic-
ulating words — whether it is natural or not — is exercised only
with the help of an instrument created by a collectivity and pro-
vided for its use; therefore to say that language (langue) gives
unity to speech (langage) is not fanciful' (CLG, 27, 11). In other
words, without a language (langue), which is a social product,
the faculty can do nothing.

2.1.2. The model of communication

These arguments for distinguishing langue from natural lan-
guage (langage) tell us neither how to relate the two nor how to
relate langue to what is left of langage when langue is, as it were,
subtracted from it. It is these questions which are addressed next;
perhaps surprisingly, a certain model of communication plays
an important role in Saussure's argument at this point. I say
surprisingly, because Saussure never explains why this model is
the appropriate one, but simply introduces it without further
justification.
This is an important point, because it seems that there are a
number of assumptions implicit in the choice of this model which
 LANGUE AND PAROLE 25

Figure 2.2

are central to his argument. The first of these is that language

is an instrument of which the primary use is in communication
(Ducrot 1968, 66; Aarsleff 1967). The second is that face-to-face
oral communication is the relevant norm (Harris 1987, 25). It
might indeed be said that the second assumption is reasonable
if his arguments that writing is a secondary, parasitic system go
through (2.2.1); though this would still leave the question of
what linguistics has to say about such secondary systems. But, as
far as I can see, the first assumption, if it is indeed one that
Saussure makes, is entirely unsupported.
Saussure's model assumes that a typical communicative ex-
change involves two people in face-to-face contact talking to each
other (fig. 2.2). Let us suppose that they engage in a simple
exchange:

(1) A: Do you have the time?

B: Six-thirty.

Precisely how is this to be described in detail? Saussure postulates

a place (module?) in A's brain (sic) 'where mental facts (concepts)
are associated with representations of the linguistic sounds
(sound images) that are used for their expression' (CLG, 28, 11).
So, for example, the concept of a cat is associated with a rep-
resentation of the sound one hears when the word 'cat' is
pronounced.
When A speaks, the following things happen:
(i) A concept (or concepts) in A's head unlocks the correspond-
ing sound image(s). This is a psychological process,
(ii) The brain transmits an impulse corresponding to the sound
image to the organs producing the sounds.
 SAUSSURE
Audition Phonation

c = concept
s = sound image

Phonation Audition
Figure 2.3

(iii) Sound waves travel to B's ear: a purely physical process,

(iv) These now activate a sound image in B's brain: a process
which Saussure describes as physiological, so presumably
(ii) is also,
(v) The sound image now unlocks the corresponding concept
in B's brain: another purely psychological process.
All of this constitutes one-half of a circuit, which is completed
when B goes on to say something. The completion of the overall
exchange is illustrated by Saussure in Figure 2.3. Saussure points
out that this circuit could be complicated, but argues that the
simple diagram includes everything essential.9 But where in all
of this are we to locate langue}
Saussure's answer seems to depend on three distinctions.
The first is between, on the one hand, what is psychological,
namely the associative relations between concepts and sound
images, and, on the other, everything else, that is the physio-
logical and physical processes. The second is between, on
the one hand, what is active, namely everything involved
in (i), (ii), and (iii), and, on the other, what is passive, that is
everything involved in (iv) and (v).10 Finally, Saussure dis-
tinguishes between what is executive (i) and what is recep-
tive (v).
At this point Saussure's argument becomes very difficult to
follow. This is in part because the text makes it natural to assume
that he is trying to locate a place for langue somewhere in the
circuit, despite the fact that if he was it would be very difficult
to make sense of the claim that langue is social. In fact what he
is doing is quite different, namely trying to explain how it comes
 LANGUE AND PAROLE 27
about that each member of a linguistic community has a similar
representation of its langue.
His argument seems to be this: So far, our account has been
too simple in that it has omitted the 'associative and coordinating
faculty... [which] plays the dominant role in the organization of
language as a system' (CLG, 29, 13). Roughly speaking, this fac-
ulty relates individual signs to other signs.11 Now, to understand
the role of this faculty we have to consider social facts. For when
a group of individuals converse with each other 'all will repro-
duce - not exactly of course, but approximately - the same signs
united with the same concepts' (ibid.). If this happens — and
Saussure of course simply asserts that it does - then the social
fact in question to be accounted for is the fact that the members
of the linguistic community in question do this.
Now, some parts of the circuit that we have been considering
cannot be responsible. The non-psychological part can be ruled
out for a start. For, to repeat the argument about nu considered
earlier, 'when we hear people speaking a language we do not
know, we perceive the sounds but remain outside the social fact
because we do not understand them' (ibid.). But neither is the
psychological part wholly responsible. The executive side is al-
ways individual, by which I take Saussure to mean that the
speaker chooses what to say. So all that is left is the receptive
side, and it is through this that 'impressions that are perceptibly
the same for all are made in the minds of speakers' (ibid.). We
can picture this social product, their common language, as
a storehousefilledby the members of a given community through
their active use of speaking (parole), a grammatical system that
has potential existence in each brain, or more specifically in the
brains of a group of individuals. For language is not complete in
any speaker; it exists perfectly only within a collectivity. (Ibid.)
So, charitably interpreted, Saussure is claiming that langue is
social because it is the product of face-to-face communicative
interchanges between members of a linguistic community, each
of whom, as a result of these interchanges, has a similar rep-
resentation of it. This accounts for the way in which a sort of
average is set up among the members of a community; moreover,
it requires us to treat the existence of the langue of a community
as something that has to be explained as the product of com-
municative exchanges between individuals. Further, on this ac-
28 SAUSSURE
count, at any given time the fundamental psychological reality
is the existence of separate but similar representations of their
langue in the brains of the members of the community. What
makes their collective existence a social fact is simply the expla-
nation of how those representations are acquired, why they are
similar, and the role they play in face-to-face communication.
There is no need to invoke the idea of a collective mind and its
representations, unless indeed this is simply a shorthand way of
talking about the representations in the minds of individuals
within the community; in other words, there is no need to posit
a supra-individual collective mind.
However, it has been argued that the image of the storehouse
suggests that as well as this 'average' conception of langue, which
only requires members of the linguistic community to have sim-
ilar representations of their language, Saussure also entertained
a different one, which does reify langue, treating it as something
independent of individual representations of it, having a poten-
tial existence in the minds of individuals but existing perfectly
only in the collectivity.
Now, whether the passage in which the image of the store-
house is introduced does commit Saussure to reifying langue
seems to me to be a moot point, though it is often taken to do
so (cf. Harris 1987, 199). Saussure does not say that a language
exists independently of all members of the linguistic community;
merely that it only exists perfectly within a collectivity. This might
only be a rather clumsy way of trying to deal with the fact that,
for instance, different speakers have different vocabularies, so
that if we wanted to construct a dictionary of their language the
result would involve idealisation in that it would include all the
items that any member of the community might use, even though
no one member would use all of them. Such idealisation would
seem to be not only harmless but essential for any serious study.
It would, of course, also be true on the average conception
that a given langue is not dependent for its existence on any
particular member of the community. Elsewhere Saussure re-
peatedly stresses the need for a community and the powerless-
ness of any individual on his own to modify langue; but, correct
or not, these points require no more than the 'average' con-
ception.
A possible source of confusion in Saussure's position might be
traced to his psychologism. The average conception calls for a
 'LANGUE' AND 'PAROLE 29
distinction between elements of a language, e.g., words and parts
of speech, and individual representations of them. But for Saus-
sure the elements of language are psychological entities, housed
in the brain, which makes it difficult, though perhaps not im-
possible, to make such a distinction. So it would perhaps be easy
to conclude that if the language is incomplete in an individual,
the remainder has to be somewhere else.
There is indeed one passage which might be cited to prove
that Saussure was committed to the reification of langue. In this
passage he points out that we can assimilate the linguistic systems
of dead languages to illustrate his claim that we can study langue
without studying its use in speech. However, this claim threatens
to undermine several of his key assumptions. Presumably this
assimilation is done on the basis of written records and without
face-to-face contact, a point which raises issues that I shall discuss
a little later (2.2.1).
Whether or not Saussure is committed to anything more than
the average conception is, of course, a key issue, to which we
shall have to return (7.3); but as an interim conclusion I would
urge that he is not, that the argument considered does no more
than explain how different speakers come to have roughly sim-
ilar representations of their langue, and that what makes the
existence of a langue a social fact is the explanation of the way
it is acquired by individuals and the role that their representa-
tions of it play in each interpreting the other's speech.
As for Saussure's claim that language is passive, it is clear that
it is really a claim about how it is learned. The individual who
assimilates it by participating in communicative interchanges
plays no active part in its assimilation. This is an important point
from his perspective. For if the individual did contribute some-
thing to the acquisition of a language, then the possibility would
arise that factors other than the purely communicative—social
ones, which are of crucial importance according to him, come
into play.
So it is important to note that the argument adduced by appeal
to the speech-circuit model is quite inadequate to show that an
individual plays no active part in the assimilation of his or her
language. Even if it is true that an acoustic image automatically
unlocks its corresponding concept, that is true only of someone
like B who has already learnt the language and for whom the
relevant associations have been established, whereas to prove his
30 SAUSSURE
point, Saussure has to show that B played no active part in
acquiring those associations. And even if it is true, as Saussure
claims, that no individual can alone change the language which
he or she speaks, this goes no way to showing that acquisition
of it is a purely passive process.
Other aspects of Saussure's theory of language require dis-
cussion, in particular his claim that it is a well-defined object.
But first I must say something about the other term of the di-
chotomy, parole.

2.1.3. Speech
It may seem odd that the discussion of the distinction between
langue and parole has taken so long to introduce the latter term,
but in this respect it is not unfaithful to Saussure's own discus-
sion. It is only in the course of the explanation of how it is that
the average conception of langue is crystallised in the brains of
members of the community that the term is introduced, almost
as an aside. After arguing that the psychological part of the
circuit is not wholly responsible for the crystallisation of lan-
guage, and that the executive component must be excluded, he
adds: 'Execution is always individual, and the individual is always
its master: I shall call the executive side speaking (parole)' (CLG,
30, 13). This suggests, given the explanation of'executive', that
the domain of speech is to be restricted to what goes on when
an acoustic image is associated with a concept, but that does not
seem to be Saussure's real view. For a little later he attributes a
much wider domain to it: 'Within the act, we should distinguish
between: (1) the combinations by which the speaker uses the
language code for expressing his own thought: and (2) the psy-
chophysical mechanism that allows him to exteriorize those com-
binations' (CLG, 31, 14). The students' notes make it clear that
by the former Saussure had in mind an individual's use of the
linguistic code to express thought, and by the latter everything
involved in speech production, including phonation. In other
words, the linguistics of parole has as its subject matter everything
in the speech circuit that is not passive; it is hardly surprising
that, if this is how it is conceived, it is heterogeneous in the
extreme.
Indeed, the subject matter of parole is even more extensive
than it seems at first sight. The brief passage on the subject gives
 LANGUE AND PAROLE 31
no idea of the extent envisaged by Saussure of the individual's
freedom when making use of a linguistic code to express
thought. It is not simply a matter of a choice between alternative
formulations of what is intuitively the same thought, such as that
between:

(2a) John kissed Mary in the garden

and

(2b) It was in the garden that John kissed Mary

in response to the questions

(3a) What did John do?

and

(3b) What happened in the garden?

The kind of freedom envisaged by Saussure is much more ex-

tensive than this, in that he seems to have seen the generation
of the sentences themselves as something which is not part of
langue but something done by individual speakers; note the sen-
tence 'belongs to speaking (parole), and not to language (languey
(CLG, 172, 124).12 Thus Saussure apparently excludes from the
object of linguistics, langue, much that modern linguistics would
include, so that its domain is extremely limited. Chomsky
comments:

He was thus quite unable to come to grips with the recursive

processes underlying sentence formation, and he appears to re-
gard sentence formation as a matter of parole rather than langue,
of free and voluntary creation rather than systematic rule. There
is no place in his scheme for 'rule-governed creativity' of the kind
involved in the ordinary everyday use of language. (1964, 23)

By contrast, the domain of parole, everything to do with langage

that remains when langue has been subtracted, is positively vast.
But though it is clear that it is vast, it is far from clear precisely
what its extent is, because it is not clear what should be included
32 SAUSSURE
in the domain of facts belonging to langage. If that were clear,
then provided it was also clear which subset of facts belonged
to langue, it would be clear what is encompassed by parole - the
facts that are left when those belonging to langue are subtracted
from those belonging to langage. However, we cannot tell from
Saussure's schematic representation of the speech circuit, which
leaves various kinds of linguistic fact unrepresented, just how
many relevant linguistic facts are in fact unrepresented on it.
For instance, those considerations relevant to the determination
of A's speech act in (1) as a request and B's as a reply are omitted,
even though B's response takes the precise form it does only
because A's remarks are recognised as a request. It is difficult
to avoid the conclusion that not only is the term parole not well
defined, but that it is hard to see how it could be. Given the
vagueness of the term 'linguistic fact', it is simply unclear what
the remainder is when we try to subtract langue from langage (cf.
fig. 2.1).
Since parole is on this account such a rag-bag, corresponding
to no natural category, the description of it as heterogeneous is,
as I said, hardly surprising. However, the proper response to
this is surely that what is called for is not concentration on the
allegedly well-defined object langue to the exclusion of every-
thing else, but an attempt to distinguish well-defined areas within
this heterogeneous domain. For example, Saussure treats what
he calls phonology, the study of the articulatory and acoustic
properties of speech, as 'an auxiliary discipline which belongs to
parole9 (CLG, 55, 33).13 That is, he tries to isolate a particular
area of parole about which he thinks it is possible to say something
systematic, particularly at the level of combinatory phonology -
the study of the constraints which the articulation of one speech
sound places on succeeding ones (CLG, 79, 51). Of course, he
may have been mistaken in supposing that there is anything
systematic to be said on this subject, but he was surely right to
try to isolate one or more areas of parole for systematic study.
The crucial question, then, seems to be not how we rigorously
define parole, for that is an enquiry which can lead nowhere, but
rather what explanatory functions parole has in Saussure's sys-
tem, and which areas within the heterogeneous domain need to
be isolated for purposes of these explanations. There seem to
be three such functions. First, parole is needed, as we saw, to
explain how the social crytallisation of langue occurs. What is in
 LANGUE AND PAROLE 33
question in this case are those systematic features of language
use which make face-to-face communication possible. Second,
parole is needed to explain how the speaker constructs a sentence;
this is a matter of his or her free choice of elements belonging
to langue. Third, as we shall see (4.5), Saussure appeals to parole
to explain how languages change: 'It is in speaking that the germ
of all change is found. Each change is launched by a certain
number of individuals before it is accepted for general use' (CLG,
138, 98). Thus parole is needed to explain not only how langue
is constituted as a stable system in a community, but also how
changes occur in it.
Now, precisely what theories are needed to perform these
functions is not a question that we can answer here — though,
as we have just seen, many would argue that the second is per-
formed by an enriched concept of langue which includes syntax.
But it is at least clear in outline that in the first case a detailed
explanation would have to draw on a theory of the articulation
and perception of speech sounds and of their functionally sig-
nificant features, that is on phonetics and phonology as well as
a theory of the constitutive features of speech acts. Whether
these could all be regarded as disciplines auxiliary to a study of
langue is a separate issue. But it is surely clear that instead of
continuing to talk about parole as an undifferentiated whole op-
posed to langue, conceptual clarity requires that one attempt to
isolate areas of study within that whole and investigate their
dependence on the study of langue. Nor does it seem that Saus-
sure could object to this - nor indeed would have, given his
treatment of phonetics (what he calls phonology) as an auxiliary
science, since it is only when this has been done that the metaphor
of langue as the platform on which everything else rests can be
cashed in. Indeed, he says as much:

In granting the science of language (langue) its correct place within

the overall study of speech (langage), I have at the same time
located linguistics itself. All other elements of speech - those that
constitute speaking (parole) - freely subordinate themselves to the
first science, and it is by virtue of this subordination that the parts
of linguistics find their natural place. (CLG, 36, 17)

It is true that the degree of subordination envisaged by him is

at times considerable. Language is likened to a symphony, which
34 SAUSSURE
exists independently of any of its performances; mistakes, if
there are any, are the fault of the musicians and not defects in
the symphony. The vocal apparatus itself is likened to a device
used to transmit Morse Code, that is to something of no intrinsic
interest in relation to the code, which could be transmitted in
many other ways.
This last analogy, indeed, suggests that strictly speaking the
nature of the instrument is of no interest at all. But that would
not be true of the first. Symphonies are written for specific in-
struments and contain detailed instructions for their use. Which
analogy is followed obviously makes a difference; as we shall see,
Saussure follows sometimes one, sometimes the other.
On this account of the Saussurean programme, it is clear why
the study of langue should be viewed as pre-eminent. It is because
in a quite literal sense it is what gives unity to the study of
language (langage); and because this is so, other areas of study,
those belonging to parole, are subordinate to it. This account is,
of course, quite different from one that advocates concentration
on the study of langue 'in and for itself, as did the editors in the
last sentence of CLG, at least if this is interpreted as a warrant
to ignore the diverse studies belonging to the field of parole (TM,
476). Moreover, it should be clear that anyone who took this
advice would be in the position of someone who had climbed a
ladder and then thrown it away. The concept of parole is essential
for Saussure's general theory of language, as is evident from the
list of explanatory functions considered above.
Earlier in the discussion Saussure pointed out that language
(langage) has both a social and an individual aspect. It should
now be clear in what respects he thinks it social and in what
respects not; everything pertaining to langue is social, whereas
parole is individual.

2.1.4. Langue as a self-contained object

Of course langue could only have the pre-eminent position Saus-
sure gives it if it is indeed a self-contained object capable of an
independent definition (cf. 2.1). But why did he think it was?
The first reason he gives seems less than convincing: 'It can be
localized in the limited segment of the speaking-circuit where
an auditory image becomes associated with a concept' (CLG, 31,
14). For there are a variety of respects in which this is not self-
 'LANGUE' AND 'PAROLE' 35

contained; acoustic images, for instance, are causally related to

other parts of the circuit. Nor is the fact that we are dealing with
the social side of speech - something institutional existing 'only
by virtue of a sort of contract signed by members of the com-
munity' (ibid.) - a guarantee of the requisite sort of independ-
ence, for many institutions are constrained by external factors.
If it is an institution, it will have to be one of a very special sort
to constitute a self-contained object, which indeed Saussure
thinks it is: 'Several features set it apart from other political,
legal, etc. institutions' (CLG, 33, 15).
What makes language special, he argues, is that it is a se-
miological system comparable to other systems of this sort, such
as writing, Morse Code, and the sign language of the deaf (7.4).
But though comparable it is the most important of such systems:

Linguistics is only a part of the general science of semiology; the

laws discovered by semiology will be applicable to linguistics, and
the latter will circumscribe a well-defined area within the mass of
anthropological facts The task of the linguist is to find out
what makes a language a special system within the mass of se-
miological data. {CLG, 33, 16)

Since the science of semiology did not exist, Saussure's move at

this point is, to say the least, surprising. There were no known
semiological laws, so there was precisely nothing to appeal to.
Moreover, the history of semiology since Saussure's time does
not inspire confidence that there are any discoverable laws at
all (7.4).
However, what Saussure is proposing is, I suggest, more mod-
est. Granted that there were no known semiological laws to ap-
peal to, it would still be possible to make a start on the
development of semiology by asking what is special about lin-
guistic signs, investigating the ways in which they form systems,
and whenever possible making comparisons with other sign sys-
tems. This last proposal would be likely to be profitable only if
future semiological investigations in other areas bore fruit. But
Saussure's reason for making it seems to be primarily method-
ological, namely that whilst comparisons with legal and political
institutions are likely to distract one from the essential features
of language, comparisons with other human sign systems are
less likely to do so. And since Saussure signally fails to make any
36 SAUSSURE
such comparisons, the point is anyway arguably not materially
important for the actual development of the argument of CLG
(7-4)-
If this is right, then the appeal to semiology amounts to little
more than an outline of work that needs to be done. In partic-
ular, the characterisation of the nature of language developed
so far is incomplete and cannot be completed until an account
has been given of the nature of linguistic signs and of the way
in which they form a system.
This account will form the subject matter of the next two
chapters, but before we turn to them we must do three things:
first, say something about Saussure's conception of a language
as a form; second, discuss the relation of speech and writing to
langue in Saussure's system; and finally, ask, once more, why he
did not take the faculty of language as the object of enquiry.

2.2. Language as a form

Earlier we saw that Saussure maintains that sounds are only
the instruments used by speakers to communicate and are of
no intrinsic interest as such, any more than is the instrument
used to transmit Morse Code. That is why sounds form the
subject matter of the auxiliary science phonology, which it
should be stressed again is not to be identified with what a
modern linguist designates by that term. What is of primary
interest to the linguist, by contrast, are not physical entities at
all, but psychological entities (acoustic images and concepts)
and the relations that obtain between them and other entities
of the same kind:

It [langue] is a system based on the mental oppositions of auditory

impressions, just as a tapestry is a work of art produced by the
visual oppositions of threads of different colors; the important
thing in analysis is the role of oppositions, not the process through
which the colors were obtained. (CLG, 56, 33)

So we must make a sharp distinction between, on the one hand,

the entities belonging to langue, acoustic images and concepts,
and the relations between them, and on the other the entities
used to communicate (sounds, graphemes, etc.).
This claim leads to another central thesis, namely that Ian-
 LANGUE AND PAROLE 37
guage is a form, not a substance: 'Linguistics then works in the
borderline where the elements of sound and thought combine;
their combination produces a form, not a substance' (CLG, 157, 113).
By this I take Saussure to mean that the significant distinctions
belonging to langue are not based on independently identifiable
distinctions in another medium, such as sound, but are, on the
contrary, ones which provide principles of classification for that
medium. In other words, the form of a language is a set of
abstract relationships which are realisable in a particular medium
or substance but which are not determined by the nature of that
medium (Lyons 1968, 56).
Now, if language is a form and not a substance, it would seem
that it is a matter of indifference which substance is utilised as
its means of expression, provided that the abstract relationships
are realisable in it. Various things Saussure himself says seem to
commit him to such a view. For example, commenting on Whit-
ney's claim that it is only through luck that we use the vocal
organs as a means of expression, he says that 'the choice was
more or less imposed by nature. But on the essential point the
American linguist is right: language is a convention, and the
nature of the sign that is agreed upon does not matter' (CLG,
26, 10). And a little later he argues that 'what is natural to man-
kind is not oral speech, but the faculty of constructing a lan-
guage, i.e. a system of distinct signs corresponding to distinct
ideas' (ibid.). So it seems to follow that not only is it a matter of
indifference which substance is used, but there is no reason why
one and the same language should not use different means of
expression for different purposes, e.g., sounds for face-to-face
communication and graphemes for writing, without either of
them having a privileged position.
It is, therefore, surprising to find that this is a conclusion which
Saussure rejects; though sounds are only the instrument of
langue, they are nevertheless an instrument which has a special
relation with it: 'Language and writing are two distinct systems
of signs; the second exists for the sole purpose of representing
the first. The linguistic object is not both the written and spoken
forms of words; the spoken forms alone constitute the object'
(CLG, 45, 23). This is to appeal to the first of the two analogies
considered earlier, namely the analogy with a symphony, which
is written for specific instruments, so that transcriptions are
clearly derivative. The other analogy, with Morse Code, which
38 SAUSSURE

Saussure Lyons
Acoustic image Expression element

[bet] [bet] 'bet

'bet'

Figure 2.4

makes the question of the nature of the instrument a subordinate

one, is at this point pushed into the background. But apart from
the fact that relegating writing to the status of a secondary par-
asitic system seems to contradict the claim that it is a matter of
indifference which kind of sign is agreed on, Saussure, when he
writes that the spoken form alone constitutes the object, also
seems to contradict his claim that acoustic images and concepts,
and the relations between them, are the primary objects of in-
terest to linguistics.
One diagnosis of what has gone wrong locates the problem
with the term 'acoustic image'. This suggests that the entities that
constitute langue are direct representations of sounds, but it is
not clear why they have to be. If they were conceived of in a
more abstract way, and called 'expression elements', following
Lyons, then it would be possible to relate them both to repre-
sentations of sounds indirectly by means of something analogous
to morpho-phonemic rules and to representations of written
words by a different set of rules. This would seem to be fully in
the spirit of the idea that language is a form, and is the position
adopted by Lyons in his admirably clear exposition of a version
of structural linguistics which draws its inspiration from Saussure
(1968, 60).14 So a particular English word which is realised (in-
directly) as a sequence of sounds by [bet], will be realised ortho-
graphically as 'bet'. Saussure's view is contrasted with Lyons's in
Figure 2.4. It is clear, however, that such a solution would not
have been acceptable to Saussure. For on this account it is dif-
ficult to see in what way the orthographic representation is par-
asitic on the phonological one, and is in fact a representation of
it. For granting that historically speech always antedated writing,
 LANGUE AND PAROLE 39
and that 'all systems of writing are demonstrably based upon
units of spoken language' (Lyons 1968, 39), it does not follow
that a written word actually represents its spoken form. And
even Saussure would have to concede - particularly as this is
part of his complaint about the tyranny of written language -
that, once established, the written language would acquire a life
of its own. Even if it started out as a parasitic system, it would
become increasingly autonomous.

2.2.1. Derrida, Saussure, and writing

Pointing out these tensions and inconsistencies in Saussure's
thought, Derrida has argued that they arise in part because Saus-
sure for metaphysical reasons had to privilege sound:

Although he recognised the necessity of putting the phonic sub-

stance between brackets ('What is essential in language, we shall
see, is foreign to the phonic character of the linguistic sign' [p.
21]. 'In its essence it [the linguistic signifier] is not at all phonic'
[p. 164]), Saussure for essential, and essentially metaphysical, rea-
sons had to privilege speech, everything that links the sign to
phone. He also speaks of the 'natural link' between thought and
voice, meaning and sound (p. 46). He even speaks of the 'thought-
sound' (p. 156). (Derrida 1981, 21)

One reason Saussure had to privilege speech is that the very

concept of a sign, which he employed because he could think of
no other,

carries within itself the necessity of privileging the phonic sub-

stance and of setting up linguistics as the 'pattern' for semiology.
Phone, in effect, is the signifying substance given to consciousness as
that which is most intimately tied to the thought of the signified
concept. From this point of view, the voice is consciousness itself.
When I speak, not only am I conscious of being present for what
I think, but I am conscious also of keeping as close as possible to
my thought, or to the 'concept', a signifier that does not fall into
the world, a signifier that I hear as soon as I emit it, that seems
to depend upon my pure and free spontaneity, requiring the use
of no instrument, no accessory, no force taken from the world.
(Ibid.)
40 SAUSSURE
But granted for the moment that Saussure did subscribe to a
metaphysics of presence, so that for him the meaning of a speak-
er's words is an expression of the speaker's thought, it is not
obvious that in a manuscript culture speech sounds are for every-
one the paradigm of a signifying substance. And it is even less
obvious that the concept of a sign carries with it the necessity of
privileging phonic substance; it is hard to see how for instance
the sign language of the deaf does that (7.4). So though Derrida
may be right that Saussure considered speech sounds to be the
paradigm of a signifying substance, he certainly did not think
that it was inevitable in a manuscript culture that everyone would
be of the same opinion, for 'the spoken word is so intimately
bound to its written image that the latter manages to usurp the
main role. People attach even more importance to the written
image of a vocal sign than to the sign itself {CLG, 45, 24). In
other words, it is not clear that Saussure thought that it was
necessary from a phenomenological point of view to consider
speech sounds as the paradigm case of a signifying substance.
Of course Derrida is right to draw attention to the fairly ob-
vious tensions in Saussure's thought, in particular that between
the view that, on the one hand, speech sounds are but one of
many possible instruments which can be used to realise a lan-
guage — the instruments as such being of no interest — and, on
the other hand, the view that there is a particularly important
connection between sounds and meaning. This difference is, of
course, reflected in the difference between the Morse Code and
symphonic models of the relationship between language and its
instruments. Moreover, he is surely right to argue that Saussure's
desubstantialised concept of the sign as a union of a 'concept'
and an acoustic image is a not very promising starting point for
a general semiology, if for no other reason than that 'it is difficult
to see how it could be extended to every sign, be it phonetic-
linguistic or not' (Derrida 1981, 22). For instance, as Derrida
points out, 'theoretical mathematics... has never been absolutely
linked with a phonetic production'; and

beyond theoretical mathematics, the development of the practical

methods of information retrieval extends the possibility of the 'mes-
sage' vastly, to the point where it is no longer the 'written' trans-
lation of a language, the transporting of a signifier which could
remain spoken in its integrity. (1976, 10)
 LANGUE AND PAROLE 41
2.2.2. Writing and the recovery of the past
A quite different reason why Saussure may have wished to priv-
ilege spoken language is suggested by Harris. It is concerned
with his desire to salvage something from the wreck of historical
linguistics. That he did wish to do this seems clear, for in contrast
to his treatment of phonology, the study of the speech sounds
utilised by a language, which he regards as an auxiliary discipline
belonging to parole, he treats phonetics, the study of the evolution
of sounds, as a 'basic part of the science of language' (CLG, 56,
33). But if it is, an acute problem arises about the evidential basis
of such a science. Clearly, much of our evidence exists only in
the form of written texts; but if writing is simply a different way
of expressing a language, then the written form will not be a
reliable guide to the spoken word. So, as Harris says, unless
writing has the status of a secondary parasitic system, which
represents langue,

he would be obliged to conclude that in the case of dead languages

a study of la langue ('the social product stored in the brain' [p.
44]) is impossible in principle. In other words most of the work
done by Indo-Europeanists of Saussure's own generation would
have to be excluded from linguistics. (1987, 43)

To which one might add: so would the work which made Saus-
sure's own reputation.
This interpretation is certainly confirmed by the reasons Saus-
sure gives for including a section on writing in a disquisition on
general linguistics. This is necessary because the linguist ought
to acquaint himself with the widest possible variety of languages.
However, 'we generally learn about languages only through writ-
ing' (CLG, 44, 23), and that is why it is necessary to consider
both its utility and its defects.
Moreover, this interpretation explains why Saussure was in-
terested in an ideal alphabetic form of writing which would make
possible a completely faithful representation of the acoustic im-
age (Harris 1987, 44).15 It also explains why Saussure thought
that the fact that we can assimilate the system underlying dead
languages does not constitute a serious counterexample to the
primacy he allots to speech sounds. For if a writing system is
parasitic on a phonic system and is a more or less accurate rep-
42 SAUSSURE
resentation of it, then we cannot reconstruct the underlying
langue without reconstructing the system of speech sounds it
utilises. So, though as a reconstruction of what motivated Saus-
sure this explanation has a more limited scope than Derrida's,
it seems to me to be the more plausible explanation, given that
Saussure did not wish to write off historical linguistics but rather
to argue for a sharp separation of synchronic and diachronic
issues.
That said, the existence of ideographic systems threatens to
undermine Saussure's position entirely. In these cases, Saussure
claims, the sign represents the whole word without representing
its internal structure, the classic example of such a system being
Chinese. However, since the ideogram may have a structure of
its own which depends on quite different principles, there seems
to be no warrant for saying that the ideogram represents the
spoken word rather than that it is a different way of representing
the same idea. The mere fact that there is a one-to-one corre-
lation between spoken and written forms does not entitle us to
say that the latter represent the former; there is such a corre-
lation in Lyons's model, for instance. Oddly enough, Saussure
admits as much: 'To a Chinese, an ideogram and a spoken word
are both symbols of an idea; to him writing is a second language,
and if two words that have the same sound are used in conver-
sation, he may resort to writing in order to express his thought'
(CLG, 48, 26). But this concession seems to be fatal to his overall
position.

2.3. The faculty of language

Saussure's claim that 'what is natural to mankind is not oral
speech, but the faculty of constructing a language' (CLG, 26, 10)
and that 'beyond the functioning of the various organs there
exists a more general faculty which governs signs and which
would be the linguistic faculty proper' (CLG, 27, 11) prompted
the question why the faculty in question should not be the object
of enquiry (2.1.1). However, discussion of this question was post-
poned after notation of the germ of a possible answer to it,
namely that the faculty cannot be exercised without the use of
a langue, which is a social product.
But granting for the purpose of the argument that langue is
indeed a cultural product, the fact that it is would be a reason
 LANGUE AND PAROLE 43
for neglecting the faculty of language completely only if it were
wholly a cultural product; that is, if the faculty made no contri-
bution at all to the structure and content of langue. Now, it is
arguable that Saussure does take care to ensure that the faculty
makes no such contribution, for, as we saw, in his account of the
way in which langue is assimilated he claims that the subject is
passive.
However, that claim fails for at least three reasons. First, as I
argued earlier, even if in a normal conversational situation the
hearer's role is a purely passive one, that is irrelevant to the
question of what is the role of a hearer learning a language from
scratch. Second, in any case, the claim that the hearer's role in
the normal speech situation is a passive one is not sustainable.
For, apart from the fact that the hearer may have to resolve
ellipsis, the words uttered may be ambiguous in a variety of ways.
Consider

(4) I will take the old men and boys to the bank.

Interpretation of this will involve resolution of a grammatical

ambiguity, 'old men and boys', and of a semantic one, 'bank',
and a decision as to whether the utterance is a prediction or an
undertaking. Third, nothing is said to rule out the possibility
that the faculty of language does indeed play a role in the ac-
quisition of language, though one of which the subject is una-
ware. Saussure could hardly object to this possibility on
grounds of principle, since he allows that speakers internalise a
representation of their langue without knowing that they do. So
it is difficult to see how Saussure could exclude the possibility
that the faculty of language makes a contribution, indeed pos-
sibly a large one, to the structure and content of langue.
It might be suggested that he is unworried by this possibility,
because for him there is no such faculty, simply a more general
sign faculty which is implicated in the use of any sign system
(Harris 1987, 19). Linguistics then would not be concerned with
such a faculty, since language is only one sign system and the
study of it is not concerned with the other system, including
writing. But, apart from the fact that to posit such a faculty
would have been to draw a very large cheque on the yet to be
developed discipline of semiology, surely questions analogous to
those we asked about the faculty of language could be asked
44 SAUSSURE
about an all-purpose sign faculty. After all, Saussure believed
that there are semiological laws, so there is nothing strange for
him in the idea that a particular semiological system is con-
strained in various respects.

2.4. How abstract is langue}

Culler points out that the contrasts to which Saussure appeals
to distinguish langue from parole do not all draw the distinction
in the same place:

... in separating langue from parole one separates the essential

from the contingent, the social from the purely individual, and
the psychological from the material By the first [criterion] la
langue is a wholly abstract and formal system; everything relating
to sound is relegated to parole since English would still be essen-
tially the same language if its units were expressed in some other
way. But clearly, by the second criterion we would have to revise
this view; the fact that Ibl is a voiced bilabial stop and /p/ a voiceless
bilabial stop is a fact about the linguistic system in that the indi-
vidual speaker cannot choose to realize the phonemes differently
if he is to continue speaking English. And by the third criterion
one would have to admit other acoustic differences to la langue,
since differences between accent and pronunciations have a psy-
chological reality for speakers of a language. (1976, 81)

These points are well taken; evidence of Saussure's own uncer-

tainty is the fact that he sometimes likens langue to Morse Code,
an abstract system implementable in many different ways, and
to an orchestra score, which is meant to be implemented in a
very specific way.
The difference between these metaphors corresponds to the
distinction between a schema and a norm drawn by Hjelmslev,
who argued that the simple dichotomy between langue and parole
drawn by Saussure should be replaced by a more complicated
set of distinctions among schema, norms, usage, and parole.

Usage is a statistical regularity: one can chart the frequency of

different pronunciations or of other uses of linguistic elements.
A speaker of a language has a certain freedom in his choice of
usage. The norm, however is not a matter of individual choice.
It is not described statistically but represented as a series of rules:
 LANGUE AND PAROLE 45
e.g. the phoneme /p/ is realized in English as a voiceless bilabial
stop. Finally the schema is the most abstract conception of struc-
ture. Here there is no reference to phonic substance. Elements
are defined in abstract relational terms: /p/ is to /b/ as I\J is to
/d/, and it is irrelevant what actual features are used to manifest
these differences. (Ibid.)

Clearly we get very different conceptions of langue depending

on whether we identify it with a schema, a schema plus a set of
norms, or a schema and both a set of norms and a set of usages.
The Morse Code analogy articulates the first of these concep-
tions, whilst that of the orchestra articulates the second. But
which is Saussure's real view? It is tempting to argue that he
does not really have to choose, since these views simply represent
descriptions of the same thing at different levels of abstraction.
But, as we shall see, an assumption Saussure makes later when
developing the theory of the sign, that it is a two-sided entity,
makes it difficult if not impossible for him to adopt this view
(3.1, 7.1). As it is, he vacillates between the conception of langue
as schema and as norm.

2.5. Summary and prospect

We have seen that to make his central distinction between langue
and parole Saussure has to distinguish three terms, natural lan-
guage - langage; a particular language - langue; and speech -
parole. Moreover, though his arguments against treating speech
sounds as the subject matter of linguistics are convincing, as is
his contention that language (langage) has both social and indi-
vidual aspects, the nature of the central dichotomy remains un-
clear in a variety of respects. One problem stems from the nature
of the definition employed, which involves division of a field of
enquiry the extent of which is assumed to be antecedently clear.
But, as we saw, this is just not so. It is unclear precisely what
totality of linguistic facts is meant to be included in langage, given
the inherent vagueness of the term linguistic fact'. So it is hardly
surprising that, whatever one takes langue to be, the 'remainder',
parole, should turn out to comprise a heterogeneous collection
of very diverse kinds of things. But surely no conclusion of
substance can follow from this; for instance, that parole does not
merit serious study. The proper conclusion is that it comprises
46 SAUSSURE
a variety of areas which may well merit serious study, and that
talk of a dichotomy is dangerously misleading if it blinds us to
that fact.
I have argued that Saussure's claim that langue is the object
of linguistics should not be undertood to entail that it and only
it should be studied. What is claimed is not this, but rather that
reconstruction of a langue is an essential prerequisite to any kind
of serious linguistic enquiry. This interpretation makes Saus-
sure's interest in historical and geographical linguistics intelli-
gible, which the first does not. As for Saussure's claim that langue
is social, that would seem not to involve any commitment to a
supra-individual mind, so it does not involve any dubious re-
ification of that sort. The only mental states there are are those
of individuals. What makes langue social is the explanation of
how members of the same linguistic community acquire similar
representations, i.e., how they come to associate systematically
the same concept with the same acoustic-image, and the role
those representations play in face-to-face communication. This
is, no doubt, somewhat cryptic as it stands; we shall return to a
more detailed discussion of this issue in Chapter 7.
Since, as I have argued, the characterisation of langue is in-
complete, because the way in which it constitutes a system has
yet to be explained, an evaluation of the utility of Saussure's
concept is not appropriate at this point. Nevertheless, there are
a number of questions about the way in which the concept was
introduced. In the first place, we are never told why the speech-
circuit model is the appropriate model to appeal to; yet there
seem to be a number of important assumptions implicit in that
appeal. Second, Saussure's arguments showing that langue is pas-
sive, that is that individuals have no active role to play in its
acquisition, seem to be quite inadequate to demonstrate that this
is so. As a result, his attempt to exclude a role for the faculty of
language in language acquisition, which would, if it existed, di-
minish the importance of social factors, is a failure. Third, it is
not easy to see what is to be gained by treating langue as a se-
miological system in the absence of any account of what such a
system contains.
Perhaps the best answer Saussure could give to this last com-
ment is to demonstrate the value of adopting the semiological
approach in his explanation of the way in which langue consti-
tutes a system. That, anyway, is what he seems to do; and it is
to that explanation that we now turn.
 LANGUAGE AS A SYSTEM
OF SIGNS
/ : Signs, arbitrariness, linearity, and change

The title 'Language as a System of Signs' was proposed by Saus-

sure himself for Part 1, Chapter 1, of CLG, but for some reason
the editors ignored his suggestion. I have used it as part of my
title because it encapsulates a central point in Saussure's theory,
namely that a language (langue) is a system of signs forming a
well-defined object which can be studied independently of the
other aspects of natural language. Thus, as was argued in the
preceding chapter, the ideas in this and later chapters of CLG
can be seen as deepening the characterisation of langue devel-
oped so far. As we shall see, there are many occasions on which
Saussure returns to points raised earlier.
Clearly, there are two central ideas to be introduced now, that
of a sign and that of a system. To develop the former, Saussure
introduces two key principles, the Arbitrary Nature of the Sign
and the Linear Nature of the Signifier. The first of these has
been the object of a great deal of discussion; the second has
occasioned much less comment despite Saussure's claim that it
is as important. As we shall see, each principle has a major role
to play in Saussure's overall argument. However, the character-
isation of them in Part 1, Chapter 1, is only partial, and Saussure
expands it in a number of important ways in later chapters. For
instance, his discussion in Part 1, Chapter 2, of the question why,
47
48 SAUSSURE
if a sign is arbitrary, an individual — indeed, a community — is
nevertheless not free to change it is an important development
of the idea of arbitrariness. Moreover, this discussion of the
reasons why languages are both resistant and vulnerable to
change introduces for the first time temporal considerations into
Saussure's account, thus raising a number of important issues
which call for further discussion.
My next chapter, Chapter 4, will describe how consideration
of the effect of time on language leads Saussure to formulate
the second major dichotomy of CLG, that between synchronic
and diachronic approaches to language, and to give a prelimi-
nary outline of the precise way in which a language constitutes
a very special kind of system, 'a system of pure values which are
determined by nothing except the momentary arrangements of
its terms' (CLG, 116, 80). Saussure's detailed description of this
system will be reserved for Chapter 5; whilst Chapter 6 will
complete the outline of his main argument by considering his
account of linguistic value and its dependence on a system.

3.1. Language is not a nomenclature

As we saw in 1.2, Saussure was opposed to nomenclaturism, the
view that language 'is a naming-process only - a list of words,
each corresponding to the thing that it names' (CLG, 97, 65).
The only element of truth in this picture, Saussure claims, is the
conception of a linguistic unit as a 'double entity' the two parts
of which are joined by association. But otherwise the view is
seriously mistaken: It assumes the existence of ideas that ante-
date words; it leaves it unclear whether a word is a vocal or a
psychological entity; and it assumes far too simple a picture of
the relation between a name and what it names.
However, as Harris points out, it is by no means clear that all
versions of nomenclaturism are subject to Saussure's objection.
So the best way to proceed at this point is not to try to evaluate
his attack in general terms, but to see what particular version of
nomenclaturism Saussure objected to, and how compelling are
his objections to it.1
The extensive manuscript note used by the editors to supple-
ment the students' notes at this point makes it clear that the main
target was comparative linguistics (Engler 2, 148F). Treating a
word as a double entity, e.g., as a signifier correlated with some-
 LANGUAGE AS A SYSTEM OF SIGNS I 49
thing else, might suggest that it is possible to trace the history
of a word of which the meaning remains fixed but which is
expressed in different ways in different languages. From this
point of view, there is nothing wrong in principle with the sug-
gestion that the Latin word arbor and the French word arbre
express the same idea using different means, since the idea ex-
pressed is not tied to any particular form of expression. But, as
we saw (1.2.), Saussure believed that this was a mistake; the
changes in time that affect the form of expression also affect the
idea expressed. For instance, the Latin crimen (accusation) be-
came our 'crime'.
In sharp contrast to the comparativists, Saussure held that
there are no language-independent meanings.2 It is true, of
course, that the kind of argument adduced which appeals to
changes of the crimen/'crime' sort is purely empirical. So even
though Saussure cites many instances of changes of form that
are also changes of meaning, what is badly needed at this point
is a theoretical justification of his position. There is no doubt
that it is his position.
Moreover, Saussure's stance on the other two issues raised —
viz., how complex is the relation between a name and what it
names, and whether signs are vocal or psychological — is closely
connected to his claim that there are no language-independent
meanings. For if there are none, then it is plausible to suppose
that the meaning which a particular item has depends in some
way on its intralinguistic relations to other items in the language
rather than on its relation to something extralinguistic, so that
if it is a name of something it is so only because of the role it
plays in the language as a whole. And if it is knowledge of a
sign's intralinguistic relations which enables members of a lin-
guistic community to engage in communication, then those re-
lations must be psychologically real. This is an assumption which
is obviously central to Saussure's speech-circuit model, which in
effect treats the relations as holding between psychological enti-
ties, which, though they are not called such in his exposition,
are signs. Hence, for him a word is a psychological entity.3
So the position that Saussure tries to establish is that a linguistic
sign is a psychological entity related systematically to other lin-
guistic signs to form a system, and furthermore one whose sig-
nifying features depend on the ways in which it is related to
those other signs. But before trying to trace his argument, it is
50 SAUSSURE

Figure 3.1

worth stressing that he does retain a relic of nomenclaturism in

the assumption that a sign is a 'double entity' in a surprisingly
uncritical way, and that one of the more difficult things he has
to do is to explain why, if each sign has a double aspect, neither
of the aspects can exist independently of the other.

J.I.I. Linguistic signs

In one fairly natural usage, if a word expresses an idea it might
be said that it is a sign of an idea. But this usage is not Saussure's.
For him, a sign does involve two things, an acoustic image and
a concept, but he does not think of the former as a sign of the
latter. On the contrary, the sign is the union of both of them,
and can be represented as in Figure 3.1. We are by now familiar
with the term 'acoustic image' which is explained here as the
'psychological imprint of the sound, the impression that it makes
on our senses' (CLG, 98, 66). This explanation troubled the ed-
itors, who worried whether it should not also include a repre-
sentation of the movements involved in articulation, presumably
to account for its role in execution as well as that in reception
(ibid.). But the point is that though Saussure's term 'acoustic
image' makes it very natural to think of it as an image, there
would seem to be no need to do so. The psychological imprint
could be quite abstract given the role it plays in the speech-circuit
model; all that is necessary is that there should be systematic
correlations between it and speech sounds, whether uttered or
heard.
In an earlier discussion we saw that the editors retained the
pair of terms 'acoustic image'/'concept' for the original expla-
nation of a sign, ignoring Saussure's instruction to use the pair
'signifier'/'signified' instead (1.3). They do introduce the latter
pair later:
 LANGUAGE AS A SYSTEM OF SIGNS I 51
I propose to retain the word sign [signe] to designate the whole
and to replace concept and sound-image respectively by signified
[signifie] and signifier [signifiant]; the last two terms have the
advantage of indicating the opposition that separated them from
each other and from the whole of which they are parts. (CLG,
99> 67)

But the only reason they give for coining the new terms is as a
corrective to the tendency to think of the acoustic image itself,
e.g., arbor, as the sign, rather than the union of it with a concept.
Inevitably, the reader is left with the impression that 'signified'
is a near-synonym for 'concept' and merely has different asso-
ciations, whereas, as we shall see, Saussure tries to elucidate the
notion of a signified in terms of the notion of a value, albeit a
very special kind of value arising from social usage (6.1). So the
suggested identification of a signified with a concept, and not
even a concept of a special kind, is, to say the least, unfortunate,
since the dangers of lapsing into the sort of nomenclaturist the-
ory that Saussure so objected to are clear.4
Arguably, the editors had no option but to retain the term
'acoustic image', since one of Saussure's central theses is that a
signifier is an acoustic image, though the 'is' in question is pre-
sumably not one of identity, since there are acoustic images which
are not signifiers. However, it might be argued that analogously
they were right to retain the term 'concept', since a signified is
a concept, albeit one of a special kind.
Certainly for the speech-circuit model to play any part in Saus-
sure's argument, a signified has to be able to fulfil the role al-
located to a concept in that model. So perhaps what we are
witnessing at this point is an attempt to reconcile two apparently
contrary tendencies in Saussure's thought, the psychologism im-
plicit in the speech-circuit model on the one hand, and the em-
phasis on the social character of language which leads him to
treat a signified as a special kind of value on the other. Perhaps
that tension is resolvable, but until it is resolved the mixture of
terminologies is undoubtedly confusing. To avoid such confu-
sion we should, at this stage, think of the new terms 'signifier'/
'signified' as theoretical terms whose content has to be explained
by reference to the role they play in a theory that has yet to be
developed, rather than as near-synonyms of the old pair 'acoustic
image'/'concept'.
52 SAUSSURE
Now, according to Saussure, linguistic signs have two 'pri-
mordial characteristics'. They are arbitrary, and their signifiers
are linear. It is to these two key theses that we now turn.

3.2. Linguistic signs are arbitrary

This is so, Saussure argues, because the bond between the sig-
nifier and the signified is arbitrary, a point which he illustrates
as follows:

The idea of 'sister' is not linked by any inner relationship to the

succession of sounds s-o-r which serves as its signifier in French;
that it could be represented equally by just any other sequence is
proved by differences among languages and by the very existence
of different languages: the signified 'ox' has as its signifier b-o-f
on one side of the border and o-k-s (Ochs) on the other. (CLG,
100, 67)

But though this explanation of what he meant is clear, it has

seemed to many not to be clearly to Saussure's purpose, since it
apparently rests on the very kind of nomenclaturist assumption
that he objected to so strongly, and so provides an illustration
of the danger of retaining the conception of the sign as a two-
sided entity. Hence Jakobson's comment that

this theory is in blatant contradiction with the most valuable and

the most fertile ideas of Saussurian linguistics. This theory would
have us believe that different languages use a variety of signifiers
to correspond to one common and unvarying signified, but it was
Saussure himself who, in his Course, correctly defended the view
that the meanings of words themselves vary from one language
to another. The scope of the word boeuf and that of the word
Ochs do not coincide. (1978, 111)

Now, the assumption which is made by the example about the

differing signifiers of 'ox', that one and the same idea can be
expressed in different languages by different means, is just the
sort of nomenclaturist assumption that Saussure objected to, so
Jakobson's point is well taken.5 But does Jakobson also wish to
maintain that the connection, which is a purely intralinguistic
one, between b-o-f and its signified 'boeuf is not arbitrary? Ap-
parently he does, for he quotes with approval Benveniste's claim
 LANGUAGE AS A SYSTEM OF SIGNS I 53
that 'the connection between the signifier and the signified is not
arbitrary; on the contrary it is necessary' (1966, 50). This is so
because 'the two have been imprinted on my mind together;
they are mutually evocative in all circumstances. There is be-
tween them such an intimate symbiosis that the concept "boeuf'
is like the soul of the acoustic image b-o-f1 (ibid., 51). It is, how-
ever, difficult to believe that Saussure would have denied this,
for later, expanding on the sense in which the sign is arbitrary,
he says:

The term should not be taken to imply that the choice is left
entirely to the speaker (we shall see below that the individual does
not have the power to change a sign in any way once it has become
established in the linguistic community); I mean that it is un-
motivated, i.e. arbitrary in that it actually has no natural connec-
tion with the signified. (CLG, 101, 68)

In other words, Saussure would agree with Benveniste that the

connection is indeed one which has been imprinted on the minds
of French-speakers, and one moreover which no individual can
change. Nevertheless, he wants to say that it is arbitrary.
Part of what he means is that another sequence of sounds,
e.g., s-o-f might have become the signifier of'boeuf. This is not
to say that as things are adult French-speakers could change
their speech habits, but it is to claim that if s-o-f had been the
signifier of 'boeuf when they learned French they would have
learned to use it as such, so that they would produce the sequence
s-o-f whenever they now produce the sequence b-0-f6 As Levi-
Strauss says, the sign is arbitrary a priori, but non-arbitrary a
posteriori. That is, there is no intrinsic connection between b-o-f
and 'boeuf but once the former becomes the signifier of the
latter in French, a French-speaker has no option but to relate
them in that way.
It must indeed be conceded that Saussure's explanation of
what he means by 'arbitrary' is fairly casual, though it should be
remembered that it was his practice to introduce key distinctions
quite sketchily before going on to develop them in detail; this
case is no exception. However, more must be said at this point
about the term 'unmotivated', used to explain his notion of
arbitrariness.
Clearly, a claim to the effect that the relation between a sig-
54 SAUSSURE
nifier and its signified is unmotivated, that is that they are not
naturally connected, excludes many different possibilities. It
would be difficult to enumerate them, but in general terms, in
all of these cases some fact about the world makes the signifier
an appropriate one for its signified.
For instance, it might be argued that the fact that cuckoos
produce a certain sort of sound makes the English word 'cuckoo'
an appropriate one for its signified. The word 'cuckoo' is, of
course, onomatopoeic; that is, there is a similarity between the
sound produced when it is pronounced and the characteristic
call of the bird it denotes. In Saussurean terminology, it is the
similarity of the signifier of the sign to the sound produced by
cuckoos which makes the signifier an appropriate one for its
signified. In other words, what makes the signifier appropriate
are facts about cuckoos and sounds and signifiers.
Onomatopoeia is one of the two prima facie counterexamples
to the principle of the Arbitrariness of the Sign discussed by
Saussure. The other involves interjections, in which one is
tempted to see, he says, 'spontaneous expressions of reality dic-
tated, so to speak, by natural forces' (CLG, 101, 69). For instance,
if when hurt I naturally make the sound [ouch], then the word
'ouch' is appropriate for the expression of that feeling because
of the similarity of the sound produced when it is pronounced
to the natural expression.
Saussure's treatment of these counterexamples is reminiscent
of Hume's treatment of a counterexample to the principle that
all ideas come from antecedent impressions, viz., that it exists
but is of small account in relation to the overall principle. Thus
Saussure argues that genuine onomatopoeic signs are rare; that
they are only approximate imitations of natural sounds; and that
they become subject to morphological and phonological evolu-
tion. His discussion of interjections is even briefer; but there
seems to be no reason to quarrel with the overall conclusion that
'onomatopoeic formations and interjections are of secondary im-
portance, and their symbolic origin is in part open to dispute'
{CLG, 102, 70).7
But what is surprising is that though it is clear that the number
of possible varieties of motivation that have to be excluded is
very great indeed, no attempt is made to enumerate them and
to consider them carefully. The reader is simply left with the
impression that the counterexamples are of little account. In-
 LANGUAGE AS A SYSTEM OF SIGNS I 55
deed, we are left thinking that all that is at issue is the question
of whether there are non-linguistic facts which make certain
signifiers appropriate for their signifieds, as in onomatopoeia;
but in fact there are many other possible kinds of motivation
which Saussure wishes to exclude.
This can easily be seen by noting that Port Royal grammarians
(1.2) could readily agree about the marginality of onomatopoeia
and interjections, and continue to maintain that linguistic cate-
gories correspond to conceptual categories — noun phrases to
substances, adjectives to modes, etc. — and that the structure of
sentences corresponds to that of thoughts.8 In other words, they
could maintain that the categories of our grammar and the struc-
tures of our sentences are motivated by the categories of logic
and the structures of our thoughts. So we need to distinguish at
least three levels of motivation: that of signifiers, as in onoma-
topoeia; categorial motivation, by which we mean the determi-
nation, partial at least, by non-linguistic facts of the lexical,
syntactical, or morphological etc. categories of a language; and
structural motivation, that is the determination, at least partial,
by non-linguistic facts of the structure of phrases and sentences.
A strong version of the autonomy of language must exclude
all three kinds of motivation. But evidently if Saussure does wish
to do this, then the arguments adduced to date go a very small
way indeed towards establishing his overall conclusion. However,
as I said, it is Saussure's custom to explain new distinctions only
partially when introducing them and then to deepen the expla-
nations later (TM, 443). So, for the moment, the only possible
verdict about the thesis that signs are unmotivated is that Saus-
sure has gone only a little way to show that this is so.

3.2.1. The scope of the principle

Before I close this discussion something must be said about the
scope of Saussure's principle of arbitrariness. Does it apply to
all linguistic signs, however complex, or only to some?
In the section under discussion Saussure gives no guidance,
so the reader is entitled to conclude that, for instance, 'twenty-
seven' is as arbitrary as are its component signs 'twenty' and
'seven'. However, Saussure argues later that we should distin-
guish between the complex sign and the simple signs of which
it is composed. Only the latter are completely arbitrary; the for-
56 SAUSSURE
mer is constructed in accordance with a pattern, 'twenty-one,
twenty-two,..., thirty-one, thirty-two,...', and for this reason
is said by Saussure to be only relatively unmotivated: 'Some signs
are absolutely arbitrary; in other words we note, not its complete
absence, but the presence of degrees of arbitrariness: the sign
may be relatively motivated' (CLG, 180, 131). This suggests that the
principle of the Arbitrariness of the Sign has universal scope but
that only those signs which are not modelled on productive pat-
terns internal to the language are completely arbitrary, the re-
mainder being relatively arbitrary.
However, the notion of relative arbitrariness muddies the
waters. The feature of 'twenty-one' that Saussure describes as
relative arbitrariness is not a less absolute form of arbitrariness
than that exhibited by 'twenty' and 'one' — after all, there is
nothing more natural about 'twenty-one' than there is about 'one
and twenty' — but an example of something quite different: lin-
guistic system. In other words, the concept of relative arbitrar-
iness rests on a confusion between the idea of motivation and
that of linguistic system (5.2.1). This is so because the root idea
of motivation is that of the determination of a linguistic feature
by something extralinguistic, whilst that of linguistic system is
that of intralinguistic determination.9 It may well be that it is
because the sign is arbitrary that linguistic system is important,
as Saussure argues: 'Everything that relates to language as a
system must, I am convinced, be approached from this view-
point, which has scarcely received the attention of linguists: the
limiting of arbitrariness' (CLG, 182, 133). But to call a system a
species of arbitrariness, relative arbitrariness, not only is con-
fusing but distracts attention away from the important question.
What features, if any, must such a system have?

3.3. The signifier is linear

The first principle was about signs, but the second is a thesis
only about signifiers. As I said, it has occasioned much less dis-
cussion. Moreover, when it has been discussed it has provoked
puzzlement, on the grounds that it is clear neither what it means
nor what its role in Saussure's system is.10 Nevertheless, Saussure
thought it as important as the first principle, since 'the whole
mechanism of language depends upon it' (CLG, 103, 70). By this
he seems to mean that without this principle his account of
 LANGUAGE AS A SYSTEM OF SIGNS I 57
the way in which a language is constituted by a system of relations
of a special kind, that is syntagmatic and associative ones, would
fall to the ground (5.3).
Now, it is important to remember at this point that Saussure's
principle concerns only signifiers, and that signifiers are acoustic
images. As such they are representations of speech sounds but
are not themselves speech sounds; just as maps may represent
a hilly terrain, which has three dimensions, without themselves
being three-dimensional, so can an acoustic image represent
things which have properties it does not.
This is important, because it is easy to suppose that Saussure
is claiming that signifiers themselves are linear, and then to object
that there is no need for them to be so even if what they represent
is linear, which is correct (Henry 1970, 89). It would therefore
be nice if one could report that Saussure does not confuse prop-
erties of the signifier with properties of what it represents. Un-
fortunately, it is not completely clear that he is not guilty of this
confusion, for he writes: 'The signifier, being auditory, is un-
folded solely in time from which it gets the following character-
istics: (a) it represents a span, and (b) the span is measurable in
a single dimension; it is a line' (CLG, 103, 70). Uncharitably, one
could take him to be saying that signifiers partake of a property
of what they represent; namely, they unfold in time.
However, it is not clear that this is what he means. For the
question which concerns him seems to be: What are the con-
straints imposed on signifiers as representations by the fact that
the medium in which they are realised is auditory? His answer
is that because the medium represented is auditory, as repre-
sentations signifiers have to represent a span of time, which has
a linear character. This interpretation is borne out by what he
says a little later. 'In contrast to visual signifiers (nautical signals,
etc.) which can offer simultaneous groupings in several dimen-
sions, auditory signifiers have at their command only the di-
mension of time. Their elements are represented in succession;
they form a chain' (ibid.). His point here seems to be clear.
Whereas a visual signal can utilise a number of different features
at the same time (shape and colour, for instance), this is not true
of an auditory signal; it can utilise only temporal features, so
that it is only these features which an auditory signifier can
represent. So, whether or not he failed to make a categorial
distinction between signifiers and what they represent, the point
58 SAUSSURE
he is making does not depend on confusing these very different
things.
The point that the principle is one about signifiers is also
important because it means that counterexamples appealing to
the properties of elements that are not signifiers are, on the face
of it, not relevant, so that arguments appealing to phonology
are beside the point. These arguments turn on the fact that
phonemes can be analysed into bundles of features; /m/ and
/n/, for instance, have the same features except for the fact that
the latter is nasal and the former is not. Thus Jakobson, in a
severe criticism of Saussure, writes:

Bally, faithful to his master's doctrine, arrived at the thesis that

it is impossible to pronounce two sounds at the same time! This
argument is a petitio principii Two phonemes cannot be emitted
simultaneously. But it is perfectly possible to emit several
distinctive features at the same time. Not only is this possible,
it is what is normally done, since phonemes are complex entities.
(1978, 99)

But if, as Jakobson himself points out, 'the phoneme... differs

from all other linguistic values in that it is not endowed with any
specific meaning' (ibid., 61) — its role being to differentiate items
which do have meaning — then it is not a signifier (which does
have a meaning, since it is associated with a signified). So the
fact that a phoneme can be further analysed into a set of dis-
tinctive features is not a counterexample to Saussure's principle,
which is a principle about signifiers (TM, 447).
This argument has been challenged by Harris. He points out
that Saussure speaks of the elements of signifiers forming a chain,
and adds that

the putative counterexample to the principle of linearity which

Saussure raises and rejects in the final paragraph is the phenom-
enon of syllabic stress. The example would scarcely make sense
if linearity were only a principle which applied to the syntagmatic
relationship between one signifiant and the next, but did not apply
to the speech chain in toto. (1987, 71)

Harris's points are well taken. However, they are not conclusive.
The expression 'their elements' could be taken to refer to pho-
nemes amongst other things, but it is certainly not necessary to
 LANGUAGE AS A SYSTEM OF SIGNS I 59
take it so, particularly since Saussure did not think that the
sounds which realise signifiers literally constitute them. Instead,
Saussure could have been referring to the elements of a complex
signifier, e.g., 'twenty-one', whose elements would be simple sig-
nifiers. And, as far as I can see, the point about syllabic stress
would constitute a putative counterexample only if the stressed
syllable were a signifier, so that the stress modified what was
signified in some way.
Whether this is right or not, the crucial question is what role
the principle of linearity plays in Saussure's system. What he
himself says is, as we saw, quite explicit, namely that the crucial
account developed later of the relations which signifiers can en-
ter into (5.3) depends on it, though unfortunately he does not
say why he thinks this is so. A reasonable conjecture at this point
would therefore be that the principle is meant to explain why
signifiers can have these, and only these, relations. Certainly
without such an explanation there is a serious lacuna in his over-
all position, since there is no theoretical justification for his claim
that signifiers can enter into only the kinds of relations that he
describes.
If I am right, his argument is that because the medium in
which signifiers are realised is auditory, they can enter into only
the kinds of relations he describes later. And whatever one thinks
of such an argument, it is quite clear that an argument is needed
at this point. This is especially clear if one assumes that Saussure
has adopted a semiological approach, so that his aim is to derive
a detailed description of the nature of a linguistic system from
fundamental principles about signs (7.4).
But as well as a justificatory role, the principle has also a meth-
odological role, as Harris points out (1987, 77). For when we are
confronted with a complex signifier such as 'one hundred and
seventy-one', all that the principle of the Arbitrariness of the
Sign can tell us is that it is arbitrary, and that so are its constituent
signifiers. But it alone cannot tell us how to segment it into
constituent signifiers; to do this another principle is needed. So,
Harris argues, if the premiss is accepted that spoken language
is articulated as a concatenated succession of linked elements,
then 'by combining both Saussurean principles linguistics is im-
mediately provided with a method of analysis, which will consist
basically of segmenting any given catenary sequence by deter-
mining which arbitrary sets of consecutive elements in the se-
60 SAUSSURE
quence could constitute single signifiants and which could not'
(ibid., 77). And indeed one account that Saussure gives of lin-
guistic analysis does fit this description.11
However, if it is clear both what the role of the principle is,
and that such a principle is needed for the development of his
theory, it is very far from clear that the principle does the job
it is intended to do. For, to anticipate Saussure's account of the
'mechanisms' of language (5.3), only one of the relations he posits
has a linear character:

In discourse, on the one hand, words acquire relations based on

the linear nature of language because they are chained together.
This rules out the possibility of pronouncing two elements si-
multaneously. ... The elements are arranged simultaneously on
the chain of speaking. Combinations supported by linearity are
syntagms. (CLG, 170, 123)

Examples he gives of syntagms range from complex words like

're-read' to sentences. But as well as syntagmatic relations, there
is another kind of important relation, which Saussure calls 'as-
sociative': 'Outside discourse, on the other hand, words acquire
relations of a different kind. Those that have something in com-
mon are associated in the memory, resulting in groups marked
by diverse relations' (CLG, 171, 123). For instance, 'painful'
might call to mind 'delightful', 'frightful', etc., so that a particular
word is 'at the point of convergence of an indefinite number of
co-ordinated terms'. And it is, as we shall see, a cardinal point
of Saussure's theory that our understanding of a particular sig-
nifier involves knowing not only what syntagmatic relations it
has, but also what associative relations it has.
At this point, Saussure's developed theory is an outright con-
tradiction of the principle of Linearity, which says not merely
that the signifier represents a span — which in itself would not
lead to contradiction — but that the span is measureable in only
a single dimension. However, as Jakobson rightly says, the de-
veloped theory insists that two dimensions or axes are relevant:

He rigorously distinguished two axes: *(i) the axis of simultaneity

(AB) which concerns relations between coexisting things, and
from which any intervention by time is excluded, and (2) the axis
of succession (CD).
 LANGUAGE AS A SYSTEM OF SIGNS I 6l

(1978, 100)

The only possible conclusion is therefore that, far from under-

pinning the account that Saussure gives of the mechanisms of
language, the principle of Linearity contradicts it, since it denies
what the developed theory affirms, namely that signifiers are
locatable in a two-dimensional conceptual space.

3.4. The sign is both immutable and mutable

The editors thought that Saussure's claim, made in Part 1, Chap-
ter 2, that the sign is both mutable and immutable was sufficiently
paradoxical to call for editorial explanation, namely that he
wanted to stress that languages change in spite of the inability
of speakers to change them (CLG, 108, 74). But his thesis was
not quite as banal as that. He wanted to argue that the semio-
logical factors which explain why languages are stable also ex-
plain why they change. In essence his explanation is that they
are both resistant to and vulnerable to change because they are
historically constituted social institutions which, being founded
on the principle of the Arbitrariness of the Sign, have no rational
basis. In this respect they are quite unlike all other institutions.
This discussion considers languages (langues) in relation to the
historical factors, operating in time, that affect them. As it de-
velops, characterisations of both the ways in which langue is social
and the sign is arbitrary are extended considerably. It is also an
important discussion because in introducing the topic of time it
leads on to the introduction of the other central Saussurean
dichotomy, that between synchronic and diachronic linguistics.
62 SAUSSURE

3.4.1. Why languages are stable

We saw earlier, when looking at Benveniste's objections to Saus-
sure's principle of the Arbitrariness of the Sign (3.2), that Saus-
sure asserts that neither the individual nor the community as a
whole has the power to change a sign once it is established. But
why is this so?
To begin with, Saussure argues, a language (langue) is never
invented by its users; it is always inherited, the product of his-
torical forces. However, since it hardly follows that what is in-
herited cannot be changed by collective agreement, we must ask
'why the historical factor of transmission dominates it entirely
and prohibits any sudden widespread change' (CLG, 105, 72).
There are, Saussure claims, four main reasons why this is so.
First, because the sign is arbitrary, there is no rational basis for
a discussion about whether or not to change it: 'There is no
reason for preferring soeur to sister, Ochs to boeuf, etc' (CLG, 106,
73). Second, there are a multiplicity of signs. One can envisage
the replacement of one writing system by another precisely be-
cause it contains some twenty to forty letters. By contrast, lin-
guistic signs are numberless. The sheer size of the system makes
it difficult to grasp, let alone change. Third, the fact that it is a
system which can be understood only by specialists means that
the masses cannot transform it, because they are unaware of it.
Finally, language is a very special kind of institution. Most in-
stitutions involve only some people some of the time, but 'in
language, on the contrary, everyone participates at all times'
(CLG, 107, 73). So, because it is very difficult to get everyone to
change a habit, particularly when there is no rational basis for
change, language is like an inert mass that is highly resistant to
individual initiative.
One further factor needs to be mentioned, Saussure argues,
and that is time. Because language is inherited and signs are
arbitrary, they rest on tradition. 'We say man and dog because
our predecessors said man and dog' (CLG, 108, 74). So solidarity
with our traditions and customs is another factor, in addition to
the inertia of the collectivity, that makes for stability. One can
see what he means. There is an evident disutility in attempts to
depart radically from established social customs. By convention,
one stops when traffic lights are red and moves again when they
change to green. An alternative convention is imaginable, viz.
 LANGUAGE AS A SYSTEM OF SIGNS I 63
stopping when they are green and moving on red. But it would
obviously be most unwise to adopt this as one's practice unless
one was sure that everyone else would do the same. Analogously,
if one wishes to be understood, to adopt perfectly conceivable
alternatives to 'man' and 'dog' has no point unless one is sure
that others will do the same.12
It is now clear why Saussure thinks that the fact that signs are
arbitrary does not imply that either an individual, by stipulation,
or a group, by collective agreement, can change a sign. Never-
theless, his arguments do not exclude change tout court', what
they do, or are meant to do, is to exclude change that comes
about by design. Though Saussure does not deny that there is
scope for individual variation, what he does claim is that there
are many factors which make it difficult for it to become estab-
lished, i.e., to affect langue. Moreover, if it does, the resulting
change is never one that was designed: 'In contrast to the false
notion that we readily fashion for ourselves about it, a language
is not a mechanism created and arranged with a view to the
concepts to be expressed' (CLG, 121, 85). But if there is scope
for individual variation, the combined weights of tradition and
the social collectivity seem to be so powerful that one might
wonder why a language subjugated to them both does not be-
come completely static. In other words, the problem now is to
explain how change is possible at all.

3.4.2. Why languages are mutable

One would expect Saussure, before answering this question, to
review the kinds of change there can be in a language and the
sorts of factors that might be responsible for them, just as he
reviewed the sorts of factors responsible for stability. But in fact
he does this in only the sketchiest of ways, if at all. After saying
that changes can take many forms, he says:

One might think that it [change] deals especially with phonetic

changes undergone by the signifier, or perhaps changes in mean-
ing which affect the signified concept. The view would be inad-
equate. Regardless of what the forces of change are, whether in
isolation or in combination, they always result in a shift in the
relationship between the signified and signifier. (CLG, 109, 75)
64 SAUSSURE
For instance, the Latin necdre (to kill) became noyer (to drown)
in French. In this case, both signifier and signified have changed,
but, Saussure seems to argue, one should not treat these as two
independent phenomena: ' . . . it is sufficient to state with respect
to the whole that the bond between the idea and the sign was
loosened, and that there was a shift in their relationship' (ibid.).
This looks very much like changing the subject from the topic
of the admittedly very various factors responsible for change, to
that of their effect, which is said to be always the same, namely
a shift in the relation between signifier and signified. But why
that is their only possible effect is never explained — though, as
we shall see later, there are a number of theoretical reasons why
Saussure has to say that signs cannot change (4.2). Clearly, it is
far from obvious, given the principle of the Arbitrariness of the
Sign, that every change in a signifier will result in a new rela-
tionship with its signified. The kind of change involved in using
s-o-f instead of b-o-fas the signifier for 'boeuf should, according
to the principle, leave that relation unchanged. So what is the
force of claiming that in such a case the relation of the new
signifier to its signified is different from that of the old one
to its?
Turning, then, to Saussure's explanation why, in spite of the
weight of the collectivity and of tradition, change nevertheless
occurs, it is perhaps unsurprising that he has not got a great
deal to say, for he has at this stage said nothing at all about the
factors responsible for change. His main point is: 'Language is
radically powerless to defend itself against the forces which from
one moment to the next are shifting the relationship between
the signified and signifier. This is one of the consequences of
the arbitrary nature of the sign' (CLG, 110, 75). The kinds of
changes that can occur in other institutions are limited precisely
because they are not completely arbitrary. Fashion, for instance,
is constrained by the shape of the human body; but there is no
analogous restraint in the case of human language. So one of
the factors which make it difficult to change languages, their
arbitrariness, at the same time renders them vulnerable to
change.
At the risk of labouring the obvious, it should be said that this
satisfactorily explains why languages are vulnerable to change
only on the assumption that the strong version of the thesis of
the Arbitrariness of the Sign discussed earlier - and inter alia of
 LANGUAGE AS A SYSTEM OF SIGNS I 65

Time

Figure 3.2

Community
of
speakers

Figure 3.3

the Autonomy of Language - has been established, which it

clearly has not been; and that even then it is only an explanation
of why they are vulnerable, not of why in fact they change.
In conclusion, Saussure argues that to take account of the
effects of the two most important factors introduced into the
argument at this point, time and the social collective, we must
represent langue as in Figure 3.2. This replaces the represen-
tation suggested by the earlier discussion, which, to take cog-
nisance of the fact that langue is a social fact, exhibits its
dependence on the linguistic community (fig. 3.3). What is in-
adequate about this representation is that it ignores historical
factors. But to do this is to ignore something of crucial impor-
tance, for
66 SAUSSURE
the thing which keeps language from being a simple convention
that can be modified at the whim of the interested parties is not
its social nature; it is rather the action of time combined with the
social force. If time is left out, the linguistic facts are incomplete
and no conclusion is possible. {CLG, 113, 78)

But now that questions about the importance of historical issues

have been raised, the time has come to see what methodological
implications they have, and it is to these that Saussure now turns.
In so doing he introduces the second major dichotomy of CLG,
that between synchronic and diachronic linguistics — a topic
which forms the subject matter of our next chapter.

3.5. Summary
Saussure was root and branch opposed to nomenclaturism, the
view that a language consists of names whose function is to label
independently identifiable objects or ideas. In opposition to this
view, he insists that a language is a system of signs, and that the
signifying features of a sign depend on its intralinguistic relations
to other signs belonging to the same system rather than on its
extralinguistic relations to pre-existing objects or ideas. Clearly,
two key ideas need development and expansion, that of a sign
and that of a system, and it is with the first of these that this
chapter was concerned.
A sign is, for Saussure, a 'two-sided entity' consisting of the
union of a signifier and a signified. A signifier is an acoustic
image, so that, for example, the signifier of the word 'cat' is not
the sound made when it is pronounced, but an image of the
sound; it is thus a psychological entity. A signified is initially
identified with the concept associated with the signifier, but this
identification, I argued, is potentially confusing; we should treat
it at best as provisional and await Saussure's considered expli-
cation of it as a value arising from a system (6.1).
One primordial principle governing signs maintains that they
are radically arbitrary. This is so because they are unmotivated.
There is nothing about the world which makes one signifier any
more appropriate for its signified than any other. Saussure dis-
cusses two apparent counterexamples, onomatopoeia and inter-
jections, and has no difficulty in showing that these are not
serious counterexamples to the principle. However, he does not
 LANGUAGE AS A SYSTEM OF SIGNS I 67
raise, let alone discuss, other counterexamples which are theo-
retically much more important, so his arguments for the prin-
ciple of the Arbitrariness of the Sign are, at this stage anyway,
quite inadequate.
A second primordial principle concerns signifiers. It maintains
that since what a signifier represents is auditory, it can only
represent a span of time, which has a linear character — hence
the name of the principle, the Linearity of the Signifier. Though
much less discussed than the principle of the Arbitrariness of
the Sign, this principle is, in Saussure's view, equally important,
since 'the whole mechanism of language depends on it' (CLG,
103, 70). He thought that this is so, I argued, because without
appeal to the principle of Linearity, his account of the way in
which a language is constituted by a system of syntagmatic and
associative relations would be powerless to answer the question
of why these and only these types of relations constitute a lan-
guage (5.3). However, I argued that the Linearity principle is
far too strong from Saussure's point of view, since it implies that
the only features that signifiers can represent are ones of tem-
poral succession. It thus excludes the very possibility of associ-
ative relations, which are non-temporal by definition.
Discussion of the question of why languages are both mutable
and immutable introduces a new and important dimension into
the discussion: time. There are, Saussure argues, many reasons
why languages should be resistant to change. Their arbitrariness
means that a debate whether to change them could have no
rational basis; they are extremely complex, and only a few ex-
perts understand them; since everyone is a language user, there
is the difficulty of getting everyone to change old habits. On top
of all this, another factor must be taken into consideration: the
effect of time. Language is inherited, so that we say 'man' and
'dog' simply because that is what our ancestors did; hence,
changes have to contend with solidarity with our traditions and
practices as well as the weight of the collectivity.
In view of all this, it might seem impossible for languages to
change, yet they do. How is this possible? Saussure's answer to
this question is much less detailed than was his answer to the
question of why languages are stable. There is no review of the
kinds of factors that might be responsible for change; instead
we get an assertion that, whatever its nature, change can only
have one kind of effect, a shift in the relationship of the signifier
68 SAUSSURE
to the signified. Moreover, Saussure mentions only one factor
which makes languages vulnerable to change: their arbitrariness.
So his explanation of why languages are vulnerable to change
is, by comparison with his explanation of why they are resistant
to it, very thin indeed.
Finally, having introduced the topic of time and stressed its
importance, Saussure is faced with the question of what theo-
retical distinctions are called for by the study of linguistic change.
This leads him to introduce the second great dichotomy, that
between synchronic and diachronic linguistics; this is the topic
of the next chapter.
 LANGUAGE AS A SYSTEM
OF SIGNS
/ / : Diachronic and synchronic linguistics

Discussion of the factors which explain the mutability and im-

mutability of language led to awareness of the importance of
time. But few people have suspected, Saussure argues, that an
adequate treatment of the effect of time calls for a radical dis-
tinction between two branches of linguistics, synchronic and
diachronic linguistics.
At first sight, this second great dichotomy of CLG is much
easier to grasp than the first, that between langue and parole, in
that it seems to involve no more than a distinction, created by
one's point of view, between studying a thing as it changes
through time and studying it at a moment in time (Amacker
^Ib* 56)- B u t though the first of the explanations Saussure
himself gives of the distinction perhaps suggests a view like this,
in the course of subsequent discussion he elaborates a distinction
which is much more complicated and which involves criteria that
are not just temporal.
It also becomes clear in the course of this elaboration why the
relatively straightforward account given above of the distinction
will not do from Saussure's point of view. For it implies that the
subject matter of the two branches of linguistics is the same thing
studied from different perspectives, whereas it is a cardinal point
of Saussure's developed theory that this is not so. Synchronic
69
70 SAUSSURE
linguistics studies langue, which is a system that is psychologically
real, whereas diachronic linguistics is concerned with relations
of succession between individual items, which speakers are un-
aware of and which are in no sense systematic. In other words,
the two branches of linguistics have quite different subject
matters:
Synchronic linguistics will be concerned with the logical and psy-
chological relations that bind together coexisting terms and form
a system in the collective mind of speakers.
Diachronic linguistics, on the contrary, will study relations that
bind together successive terms not perceived by the collective
mind but substituted for each other without forming a system.
(CLG, 140, 100)

This conception of diachronic linguistics throws down the gaunt-

let to nineteenth-century historical linguists by asserting that
insofar as they are concerned only with historical questions, they
are not concerned with language (langue) at all — hence Saussure's
paradoxical dictum that 'there is no such thing as "historical
grammar"; the discipline so labelled is really only diachronic
linguistics' (CLG, 185, 134). In other words, Saussure claims,
historical linguistics has nothing to do with grammar, which is
the subject matter of synchronic linguistics; that of historical
linguistics is quite other.
At first sight this dictum seems to be a root-and-branch rejec-
tion of the claims of historical linguistics; and this is indeed so,
in the sense that Saussure believed that the methodologies and
presuppositions of nineteenth-century historical linguistics were
fundamentally flawed. But his critique of historical linguistics,
though unquestionably severe, was not meant to be wholly neg-
ative or to lead to the conclusion that an interest in historical
and comparative questions is improper. On the contrary, as well
as severely criticising the existing methodology and assumptions
of historical linguistics, he attempted to develop a methodology
for investigating historical and comparative questions which was
not vulnerable to the objections levelled against current prac-
tices. Indeed, the proposed new methodology for diachronic
linguistics — which should not be confused with historical lin-
guistics as it was currently practised — seems to rest on a plausible
analysis of how one should analyse and explain change in some-
thing. This involves three steps:
 LANGUAGE AS A SYSTEM OF SIGNS II 71
(i) Describing its state at two different times t and t'
(ii) Comparing the two states to see in what respects the later
state at t' differs from the earlier state at t
(iii) Identifying the causes of the differences between the two
states
This approach would, of course, make the resolution of com-
parative and historical questions depend on synchronic analyses,
since, applied to a language, step (i) of the analysis would call
for two synchronic descriptions of different states of that lan-
guage. Hence, synchronic questions are logically prior, so that
without a satisfactory answer to them diachronic linguistics can-
not even get off the ground. However, Saussure claims, the
actual practice of the historical linguist has been like that of an
artist who tries to produce a panorama of the Alps by viewing
them from different peaks at the same time:

The same applies to language; the linguist can neither describe

it nor draw up standards of usage except by concentrating on one
state. When he follows the evolution of the language, he resem-
bles the moving observer who goes from one peak of the Jura
to another in order to record the shifts in perspective. (CLG,
117, 82)

But, as we shall see, though Saussure developed a radical critique

of historical linguistics as it was then practised, his attempt to
develop an alternative methodology was inhibited by a number
of theoretical assumptions which make it difficult to make sense
of the idea that a language can change at all.

4.1. A science of pure values

Though consideration of the effects of time is Saussure's starting
point, he concedes that many disciplines which take account of
it do not need to split themselves up into two branches; astron-
omy, for instance, does not consist of synchronic astronomy and
diachronic astronomy. In fact it is only sciences which are con-
cerned with values that need to make a rigorous distinction

between (1) the axis of simultaneities (AB), which stands for the
relations of coexisting things and from which the intervention of
time is excluded; and (2) the axis of successions (CD), on which only
72 SAUSSURE
one thing can be considered at a time but upon which are located
all the things on the first axis together with their changes.

D
(CLG, 115,80)

The reason why a concern with values calls for such a distinction
is, according to Saussure's own notes, because values always im-
ply a system (Engler 2, 178). The implication is that values can
be studied only on the axis of simultaneities because there is
nothing systematic about change in their case.
Nevertheless, Saussure concedes that some values can be stud-
ied on both axes; for instance, since the value of a plot of land
is related to its productivity, we can trace its value through time
provided that 'we remember that its value depends at each mo-
ment upon a system of coexisting values' (CLG, 116, 80). Saus-
sure's thought here seems to be that it is not unreasonable to
postulate some relationship between changes in productivity and
changes in value, so that if these do not form a system they are
nevertheless systematic.
But even if this is correct, the case of linguistics is different.
What makes it different from economics is, Saussure argues, the
fact that whilst economic values are not purely arbitrary, lin-
guistic values are. Because this is so, language is a system of pure
values 'determined by nothing except the momentary arrange-
ment of its terms' (ibid.).1 Moreover, the sheer complexity of
language, one of the facts mentioned to explain its relative sta-
bility (3.4.1), compels one to study it separately on each of the
two coordinates distinguished above.
After considering various names for these two different stud-
ies, Saussure opts for 'synchronic linguistics' as the name of the
study which describes linguistic phenomena on the axis of si-
multaneities, and 'diachronic linguistics' for the other study,
 LANGUAGE AS A SYSTEM OF SIGNS II 73
which is concerned with the evolution of languages. But these
explanations are, to say the least, underdetermined. Though it
is clear that synchronic linguistics is concerned with systems of
values in ways in which diachronic linguistics is not, and that the
latter is concerned with evolution, not simply with change, whilst
the former is not, the precise differences between the subject
matter and methodology of the two sciences are at this point
very unclear. Moreover, since the two axes cross, one could be
forgiven for assuming that what we are dealing with is one and
the same thing studied from two different points of view, which
is, as we saw, certainly not what Saussure believes (Harris 1987,
89).
Before moving on to examples that Saussure gives to illustrate
the differences between the two sciences, three aspects of his
argument so far are worthy of comment. First, the claim that
linguistics differs from other sciences concerned with values by
being concerned with pure values rests on a much stronger ver-
sion of the autonomy of language than the arguments about the
arbitrariness of the sign have produced to date (3.2). For whilst
we are left to glean what exactly is the difference between a pure
value and other values, the root idea seems to be that whereas
the latter are at least partially determined by factors external to
the system to which they belong, the former is not. So if Saus-
sure's account of the linguistic sign is correct, the values asso-
ciated with it would be pure ones precisely because the sign is
radically arbitrary, that is unmotivated (3.2). However, as we
saw, Saussure's argument to show that linguistic signs are radi-
cally arbitrary fails to show that this is indeed so.
Second, it is not immediately clear why, even if the values in
question are pure ones, Saussure should not concede that lin-
guistics can study them 'historically', at least to the extent that
economics can, by comparing the synchronically determined
value of something at one time with its synchronically deter-
mined value at another; that is, by employing the method of
analysing and explaining change outlined above.
The final point is connected with this: that in differentiating
the historical study of economic values from that of linguistic
values, Saussure comes very close to maintaining that there is
no such thing as the history of changes of the latter, and a fortiori
no such thing as the history of changes in a language. For if
linguistic values are determined only by the momentary arrange-
74 SAUSSURE
ment of terms, there cannot be any systematic factors which
explain why one arrangement succeeds another. If there were,
then linguistic values would be determined by more than the
momentary arrangements in question. But even if such a view
is ultimately defensible, it is surely surprising.

4.2. Differences between synchronic and

diachronic facts
One important difference between synchronic and diachronic
facts is that the succession of the facts of language 'in time does
not exist insofar as the speaker is concerned. He is confronted
with a state' (CLG, 171, 81). In other words, whilst the current
state of their language is psychologically real for speakers, its
history is not. For instance, as Culler points out, the French noun
pas (step) and the negative adverb pas (not) have a common
origin, but

this is irrelevant to a description of modern French, where the

two words function in totally different ways and must be treated
as distinct signs. It makes no difference to modern French
whether these two signs were once, as is in fact the case, a single
sign, or whether they were once totally distinct signs whose dif-
ferent signifiers have become similar through sound changes.
(1976, 37)

So it is the present state of the language which determines for

the speaker whether there are one or two signs to be considered,
not the history of the signs themselves, of which most speakers
can be presumed to know nothing. We must distinguish those
facts which are psychologically real for speakers and which de-
termine values from those which are not and do not.
An example Saussure uses to illustrate these and other dif-
ferences between the two species of fact is the following. In
modern English there are a number of nouns of which the plural
is marked by a specific vowel change - 'foot:feet', 'tooth:teeth',
'goose:geese', etc. That is a fact about contemporary English,
which is psychologically real for speakers in that they will rec-
ognise the second member of each of the pairs above as the
plural of the first member. But the history of how these nouns
 LANGUAGE AS A SYSTEM OF SIGNS II 75
came to mark their plurals in this way is, Saussure points out,
very complicated, involving three previous stages:

Stage 1 Stage 2 Stage 3

foLfoti foUfeti fit.'fet
top.topi top.tepi top.tep
gos:gosi gos:gesi gos:ges

In the first of these stages, Anglo-Saxon, the plural was marked

by a final i, a claim which is part of a synchronic description of
that language state. Then, as a result of a phonetic change (not
limited to this class of nouns; it occurred whenever i followed a
stressed syllable [Culler 1976, 43]), the forms listed in Stage 2
evolved. In this stage, as in the first, the plural is still marked by
a final i, but it is also marked by the contrast between 0 and e.
Then another phonetic change occurred that led to the dropping
of the final i (a change which, again, was not restricted to these
cases). The result of this second change was Stage 3, in which
there is only one way of marking the plural, viz. by means of
the contrast between 0 and e. Our modern forms result from
Stage 3 as a result of the great vowel shift.
Now, Saussure argues, not only is the evolution of the modern
form complicated, but it is necessary when describing that evo-
lution to distinguish sharply descriptions of how the plural is
formed in any given state, which are synchronic descriptions,
from descriptions of the factors which have led to the evolution
of one form from another:

The relation between a singular and a plural, whatever the forms

may be, can be expressed at each moment by a horizontal axis:
Period A

Period B

Whatever facts have brought about passage from one form to

another should be placed along a vertical axis, giving the overall
picture:
76 SAUSSURE
•• • Period A

Period B

(CLG, 120, 84)

Moreover, Saussure argues, the diachronic facts responsible for

the transition from Stage 1 to Stage 3 were not directed to bring-
ing that transition about. The phonetic changes involved in the
transition from Stage 1 to Stage 2, for instance, had nothing
specifically to do with the plural, since they involved all cases in
which an i followed a stressed syllable. It just so happened that
as a result the language contained forms which marked the plural
in two different ways, both by the addition of i and by the contrast
between 0 and e.
This point illustrates one of Saussure's crucial principles about
change, namely that changes never modify language states directly.
That is, they have effects on certain items, which may then affect
the system, but if the system is modified it is always modified
indirectly. A change in the mass of one of the planets is an
example of an isolated event which 'would throw the whole sys-
tem out of equilibrium' {CLG, 121, 85). An isolated change occurs
which as such is not a change in the system; but indirectly the
system is affected. Analogously, the dropping of the final i which
led to Stage 3 is an example of a change in an element in a system
which indirectly affects the whole system, i.e., leaves only one
way of signalling the plural in the nouns in question. Because
changes do not affect the system itself, but only one or some of
its elements, they can, Saussure concludes, be studied only out-
side the system.
But if a language state is the result of changes which were not
directed to bringing it about, then it must be fortuitous; 'In a
fortuitous state (fdt.fet), speakers took advantage of an existing
difference and made it signal the distinction between singular
and plural; fot.fet is no better for this purpose than fotifoti.
In each state the mind infiltrated a given substance and breathed
 LANGUAGE AS A SYSTEM OF SIGNS II 77
life into it' (CLG, 122, 85). It follows, Saussure claims, that syn-
chronic and diachronic facts are very different:

Diachronic facts Synchronic facts

Not concerned with values
Not intentional;
the systematic
changes produced
are not designed
Involve only one term; the
new one, fit, replaces the old,
foti
Not systematic
Concerned with values
Speakers 'breathe life' into
differences2
Involve two terms; the oppo-
sition between fbt and fit ex-
presses the plural
Systematic

Strictly speaking, of course, change may involve more than one

term. What is crucial from Saussure's point of view is that it is
not systematic, so that the terms involved are in that sense iso-
lated terms.
Saussure's claim that the values of a sign in a particular lan-
guage state do not depend on facts about its history is one that
has come to be widely accepted. So too has the first of the ar-
guments he uses, namely the epistemological argument that facts
which are unknown to speakers cannot determine the signifi-
cance of what they say. However, since speakers are often aware
of changes in their language, the argument is not strong enough
to rule out all historical influences; for that, an additional argu-
ment is needed. Further, Saussure's principle that changes never
modify language states directly seems to exclude dogmatically
the possibility of factors present in a given state precipitating a
change which results in a new state. If this was indeed possible,
then the successive states would have a more intimate relation-
ship than the merely contingent connection between successive
states which is all that is possible on Saussure's own account. As
things stand, because of his view that change is neither precip-
itated by, nor directly affects, a language state, Saussure's po-
sition is one which makes it difficult to make sense of the idea
that languages change at all. For if successive states are not re-
lated to each other systematically, then it is difficult to see what
makes them states of the same language rather than simply dif-
ferent language states (4.3).
78 SAUSSURE
But before considering these issues in more detail, we must
first consider a powerful and persuasive analogy which Saussure
develops to illustrate the contrast and the differences, meth-
odological and otherwise, that he sees between the two branches
of linguistics.

4.3. Language and the game of chess

Many different and conflicting metaphors and analogies are em-
ployed in CLG to illuminate the nature of language. Perhaps the
most influential is one of three analogies developed by Saussure
to illustrate the distinction between synchronic and diachronic
linguistics: the comparison with a game of chess.3
The comparison is, Saussure says, the most illuminating, be-
cause in both cases we are confronted with systems of values: 'A
game of chess is like an artificial realization of what language
offers in a natural form' (CLG, 125, 88). He does not expand
on this claim, but presumably what he has in mind is the fact
that to learn how to play chess someone has to learn what the
point of the game is, what the relative weights of the pieces are,
and what their legitimate moves are. Moreover, none of these
things would seem to be determined by external exigencies or
designed to achieve an ulterior purpose; they are internal to the
game in the sense that they depend on the nature of the game
itself and nothing else. So if the pieces in a game have a value,
or weight, they would seem to be paradigm examples of pure
values (4.1).
There are three main points of comparison seen by Saussure
between chess and a language. First, just as in a given state of
the game the value of a piece depends on its position on the
board, and a fortiori on its relations to the other pieces, so in a
given language state the value of a word depends on its relations
to other words in that state. Second, since any given state is
momentary, the value of a piece varies from state to state; the
same is true of language states and words. Third, in a chess game
only one piece has to be moved to pass from one state to another.
This is a strict counterpart of the phenomena studied by dia-
chronic linguistics in a number of. respects: In language too,
change affects only 'isolated elements'. Nevertheless, in both
cases the move has repercussions for the whole system. In chess
an actual move is part of neither the preceding nor the sue-
 LANGUAGE AS A SYSTEM OF SIGNS II 79
ceeding system; it links them. Only states matter: ' . . . the route
used in arriving there makes absolutely no difference; one who
has followed the entire match has no advantage over the curious
party who comes up at a critical time to inspect the state of the
game' (CLG, 127, 89). The same is true of a language: 'Speaking
(parole) operates only on a language state, and the changes that
intervene between states have no place in either state' (ibid.).
Of course no analogy is perfect, and there is, Saussure con-
cedes, one important respect in which a game of chess differs
from a language. For whereas the moves made in chess are
designed to bring about the resulting states, linguistic changes
are not so designed, so that the states that result are purely
fortuitous. But this makes the need for a complete distinction
between synchrony and diachrony in the case of language even
more urgent: Tor if diachronic facts cannot be reduced to the
synchronic system which they condition when the change is in-
tentional, all the more will they resist when they set a blind force
against the organization of a system of signs' (ibid.). But this
apart, the analogy is in Saussure's view exact.
There might indeed also seem to be another important dif-
ference, namely that in a game of chess, even if the values of
pieces change from move to move, nevertheless the same rules
of the game remain in force in succeeding states - this is what
makes the successive states ones belonging to the same game.
Thus, however powerful a piece it is in a given state, a bishop
can move only along the diagonal. So some part of the value of
a piece in successive states remains constant. But on Saussure's
own account it is hard to see what could be analogous in the case
of a language to the rules of chess; that is, it is hard to see what
could be operative in a similar way in successive states of the
language. And because this is so, it is difficult to see what reason
there is on Saussure's account for treating successive states as
ones of the same language (4.2).
In the course of making his second point, that the value of a
piece changes from state to state, Saussure tries to meet this
objection: 'Rules that are agreed on once and for all exist in
language too; they are the constant principles of semiology'
(CLG, 126, 88). But apart from the fact that he once again draws
a blank cheque on a non-existent science, the comparison is
clearly wrong. The rules of semiology, if they existed, would
constrain all sign systems; what would correspond to them in
the case of chess would be not its own rules but the rules of any
80 SAUSSURE
possible game. So even if there are rules of semiology, they do
not stand to particular language states as do the rules of chess
to successive states of a game. This disanalogy is, as I have urged,
serious, because it leaves us with no account of how we identify
successive states as states of the same language, or indeed of how
we decide what to include within them or to exclude; in other
words, what to count as one state4 (Harris 1987, 93).
There are various other disanalogies or awkwardnesses in the
comparison, of which two are important. First, it is not really
true, as Saussure claims, that in' chess each state of the game is
independent of all antecedent states. Various things will tell the
casual observer that a game has been going on for a long or
short time; for example, missing pieces and the placement of
others on the board. In other words, the history of the game,
though obviously not recoverable in detail, will not be completely
opaque.5 Second, and perhaps more important, the acute ob-
server will be able to perceive various strategic possibilities in a
given state and speculate intelligently about what moves will
come next.
Perhaps Saussure's answer to this point would be that the fact
that this is so is simply a consequence of the fact that in chess
the players intend to produce the resulting states, and that this
is just another respect in which a game of chess and a language
diverge. However, such a response, as well as underscoring the
extent of the differences that flow from the admitted one,6 would
further emphasise the point that there is a total lack of connec-
tion on Saussure's account between successive language states.
The third disanalogy is that whilst there is no theoretical prob-
lem in the case of chess in distinguishing on a temporal contin-
uum between successive states and the moves that produce them,
there is a problem in distinguishing between successive language
states and the changes that produce them. In the following, 5,
Sf, etc. are states, and M, M', etc. moves in a game of chess.

AT
 LANGUAGE AS A SYSTEM OF SIGNS II 8l

The different states may of course last for differing periods of

time - as long in fact as the player whose turn it is takes to make
his next move. But the moment he makes his move a new state
comes into being, and the time taken to make the move has no
significance, so that the transition from one state to another is
in effect instantaneous. Hence, at any given time the game is in
some determinate state or other, and it is, moreover, clear what
belongs to that state and what does not. Now, if the analogy with
a language were exact, then it too could be analysed neatly into
successive language states, and the time taken to effect the tran-
sition - to make the linguistic move, as it were - could be ignored.
But in fact transitional periods may last a considerable time,
during which there is no uniform practice in the area subject to
change, e.g., the way in which the plural is marked. So what
should be said about such periods?
We shall return to this issue later; the point to note now is
that the model of change suggested by the analogy with chess is
a much more complicated one than the original one involving
only the axes of succession and simultaneity.

4.4. The two linguistics

Since, on Saussure's view, synchronic and diachronic linguistics
have different objects, it is not surprising that he should maintain
that they have different methodologies. In particular he stresses
two major differences, and in the course of doing so further
deepens the characterisation of the terms 'synchronic' and
'diachronic' in ways which underscore the point made at the end
of the preceding section: that the original model involving the
axes of simultaneity and succession is too simple for his purposes.
The first major difference is the evidential basis of the two
branches of linguistics. Synchronic linguistics is concerned only
with what is psychologically real for the members of a linguistic
community, presumably at a given time, so that 'to know just to
what extent a thing is a reality, it is necessary and sufficient to
determine to what extent it exists in the minds of speakers' (CLG,
128, 90). But in diachronic linguistics there is no such restriction.
It has two perspectives: 'One of these, the prospective, follows the
82 SAUSSURE
course of time; the other, the retrospective, goes back in time'
(ibid.). In other words, diachronic linguistics is in no way re-
stricted to facts which are psychologically real for a given com-
munity of speakers.
The second big difference is that synchronic linguistics is con-
cerned only with the 'totality of facts corresponding to each
language; separation will go as far as dialects and subdialects
when necessary' (ibid.). In other words, a synchronic study is
concerned not with all simultaneously existing facts but only with
subsets of these, namely those belonging to one language. Thus,
Saussure says, strictly speaking we should talk not of synchronic
linguistics but of idiosynchronic linguistics.
By contrast, diachronic linguistics is not similarly restricted;
'the terms that it studies do not necessarily belong to the same
language (compare Proto-Indo-European *esti, Greek esti, Ger-
man ist, and French est)' (CLG, 129, 90). Indeed, as Saussure's
example makes clear, not only is diachronic linguistics not con-
fined to one language, it is not even necessary for the terms
compared to be in use at different times, as is clear from the
reference to German ist and French est. This is so because 'to
justify the associating of two forms, it is enough to show that
they are connected by a historical bond, however indirect it may
be' {CLG, 129, 91). So the fact that a modern German word and
a modern French word have a common origin is a diachronic
fact.
It should now be clear just how inadequate the original model
of the distinction between the two linguistics, with its two tem-
poral axes, has become. For synchronic linguistics is concerned
not with all simultaneously existing facts but only with what is
idiosynchronic, whilst items which are of interest to diachronic
linguistics may be in use at the same time.
Moreover, if Saussure's explanations of the terms of the di-
chotomy are to be applicable, he must have available some in-
dependent characterisation of a language. Otherwise, it is hard
to see how, when asked what synchronic linguistics studies, he
could avoid giving the answer 'The state of a language'. But
when asked how one identifies this, the only answer hinted at
in these passages is that it is something that is psychologically
real at a given time for a community of speakers. However, since
there may be many different things which have this status —
 LANGUAGE AS A SYSTEM OF SIGNS II 83
imagine a bilingual community, for instance - the explanation
will come to a premature full stop at this point unless we are
given a characterisation of what it is for a language to be a shared
language of the community.

4.4.1. No panchronic viewpoint

Given the differences in both subject matter and methodology
of the two branches of linguistics, it is hardly surprising that
Saussure concludes that there is no panchronic viewpoint, at least
as far as what he calls 'concrete' entities are concerned. Concrete
entities are the signs that constitute a language, and they are,
according to Saussure, 'not abstractions but real objects' (CLG,
144, 102).7 For instance, if we take a word, the French word
chose, for instance, there is no point of view which combines both
the diachronic perspective, in which 'it stands in opposition to
the Latin word from which it derives, causa', and the synchronic
perspective, in which 'it stands in opposition to every word that
might be associated with it in Modern French' {CLG, 135, 95).
The sounds produced when the word is spoken can of course
be studied from a panchronic viewpoint, but they have no lin-
guistic value - a claim which follows from what Saussure said
about the French word nu (2.1).
Saussure's denial of a panchronic viewpoint was, as we saw,
qualified; there is no such viewpoint as far as concrete entities
are concerned. Does that mean that in the case of what are in
Saussure's view abstract entities (cases and parts of speech, for
instance) there is a panchronic viewpoint? Clearly not, for 'ab-
stract entities are always based, in the last analysis, on concrete enti-
ties' (CLG, 190, 138). Instead, what Saussure has in mind
when he admits the existence of a panchronic viewpoint are
the unchanging principles of semiology: 'In linguistics as in
chess there are rules that outlive all events. But they are gen-
eral principles existing independently of concrete facts' (CLG,
135, 95). For example, on Saussure's account any language
must consist of a number of signs which stand in both syntag-
matic and associative relation to each other (5.3). That is true
of all languages at all times. But, his point seems to be, this
tells us nothing at all about what signs there actually are in
any given language.
84 SAUSSURE

4.5. The second bifurcation

The introduction of the dichotomy between synchronic and
diachronic linguistics leads us to the second great bifurcation.
The first was, of course, that between langue and parole. But how
are the two dichotomies related? A diagram is introduced to
illustrate 'the rational form that linguistic study should take/

j Synchrony
( Language <
(Human) Speech < (^ Diachrony
(^ Speaking
(CLG, 139, 98)
But the diagram is something of a puzzle. It suggests that there
could be both a synchronic and a diachronic study of langue,
whilst parole itself is left completely out in the cold as the subject
matter of neither branch of linguistics.
It is even harder to avoid the conclusion that as a summary
of the preceding discussion the diagram is quite misleading when
it is viewed in the light of the remark, on the same page, that
'everything diachronic in language is diachronic only by virtue
of speaking (parole)" (ibid.). This is so, Saussure argues, because
the germ of all change is to be found in parole'.

Each change is launched by a certain number of individuals before

it is accepted for general use. Modern German uses ich war, wir
waren, whereas until the sixteenth century the conjugation was
ich was, wir waren (cf. English / was, we were). How did the sub-
stitution of war for was come about? Some speakers, influenced by
waren, created war through analogy; this was a fact of speaking
(parole); the new form, repeated many times and accepted by the
community, became a fact of language. (Ibid.)

Interestingly, this account describes change on the model out-

lined at the beginning of this chapter. There is a synchronic
description of the initial language state, sixteenth-century Ger-
man, and one of the state that succeeded it, modern German.
And there is a description of the mechanism by which the change
occurred. Seizing on an analogy, some speakers produced a var-
iant which was, as such, of no intrinsic interest. But when the
 LANGUAGE AS A SYSTEM OF SIGNS II 85
variant became the norm, so to speak, it became part of the
language, with the result that war was substituted for was. Of
course, not all variants are successful, and as long as 'they remain
individual, they may be ignored, for we are studying language'
(ibid.). Thus, on this account a diachronic study incorporates a
number of synchronic descriptions as well as an account of the
way in which they are related by facts of speech; that is, a dia-
chronic study is not just a study of langue, it is a study also of
parole. This is necessary, because it is only by reference to the
latter that an account of the mechanism of change can be given.
So it is hard to see how parole can be excluded from the subject
matter of diachronic linguistics.
Moreover, this example of Saussure's highlights certain ten-
sions in his overall account. For the fact that war was substituted
for was by analogy strongly suggests that the state of German
itself plays a part in the account of change. There existed a
potential for that change in the shape of an imitable model:
'Analogy supposes a model and its regular imitation. An analog-
ical form is form made on the model of one or more other forms in
accordance with a definite rule' (CLG, 221, 161). But if language
states have such a potential, in the form of so-called productive
forms, it seems hard to maintain that successive language states
always have the strict independence of each other that Saussure's
theory requires.8
Saussure's implied answer to this objection is characteristically
bold: He denies that in such cases there is any change. On the
contrary, what we have is a case of creation of a new form. One
proof that this is properly described as creation rather than
change is the fact that the old form remains unchanged, and,
for a time anyway, the two subsist side by side.
But this response, taken seriously, might well force him to
maintain that there is never any change; for in the only other
kind of change he recognises, phonetic change, it is perfectly
possible for individual variants of a standard form to coexist
with it before ultimately replacing it. The principle that a system
can change only if some part of it has changed would seem to
be far too strong if cases in which variants or analogical creations,
which initially coexist with original forms but ultimately replace
them, are not counted as examples of changes in the original
forms.
86 SAUSSURE

4.6. Summary and prospect

A central question raised by our discussion is which of the two
conceptions of change developed by Saussure is the preferred
one. As we saw, the model of change implicit in the chess analogy
is more complicated than the original one involving only the
intersection of the two temporal axes (4.1). Moreover, it became
clear — in the light of Saussure's claim that synchronic linguistics
is, strictly speaking, concerned with what is idiosynchronic, whilst
usages in existence at the same time can be of interest to dia-
chronic linguistics — that the distinction between a synchronic
and a diachronic study involves more than the temporal consid-
erations involved in the original model. This suggests that the
model implicit in the chess analogy is the preferred one. One
advantage of the latter model is that it certainly allows one to
take into consideration the fact that language states may be of
longer or shorter durations, for players make their moves more
rapidly at some times than at others. As Saussure says,

It is possible for a language to change hardly at all over a long

span and then to undergo radical transformations within a few
years. Of two languages that exist side by side during a given
period, one may evolve drastically and the other practically not
at all. (CLG, 142, 101)

Even so, I argued that the chess model highlights a serious prob-
lem. In chess, the moment a player makes his move a new state
of the board comes into being, and the time taken actually to
make the move has no significance. Hence, the transition from
one state to another is in effect instantaneous, so that at any
given time the game is in some determinate state or other. Now,
if the analogy with a language were exact, then the language
could be analysed neatly into successive language states, and the
time taken to effect a transition could be ignored. But clearly
this is not so, since a transitional period may last for a consid-
erable time, during which there is no uniform practice in the
area subject to change.
Saussure's implicit solution to the problem posed by the dis-
analogy is clear. In the case of two languages, one of which is
evolving rapidly and the other not, 'study would have to be
diachronic in the former instance, synchronic in the latter' (ibid.).
 LANGUAGE AS A SYSTEM OF SIGNS II 87
In other words, successive language states, which can be studied
synchronically, do not necessarily succeed each other immedi-
ately; there may intervene periods of change which can only be
the subject of a diachronic study. Reverting to the chess analogy,
this would be to treat the time taken to make a move as of
significance, so that whilst the move is being made the game is
not in any particular state at all (Harris 1987, 105). But the
implications of what is said at this point are, to say the least,
difficult to reconcile with Saussure's overall position. For surely
if there are times at which a synchronic description of a language
is not possible, then it is hard to see what a description of the
langue internalised by its speakers could be. But one of Saussure's
central claims is that without langue there can be no parole. So if
no synchronic description of a language is possible, it is hard to
see how, on his principles, there could be a diachronic one either.
Further, Saussure writes as though the conception of a stable
state is, at best, a methodological fiction: ' . . . since language
changes somewhat in spite of everything, studying a language-
state means in practice disregarding changes of little importance,
just as mathematicians disregard infinitesimal quantities in cer-
tain calculations, such as logarithms' (CLG, 142, 101). But fiction
or not, the assumption that there are such states, which, more-
over, form the subject matter of synchronic linguistics, seems to
be central to Saussure's thought. However, it is far from clear
that he succeeded in developing a satisfactory model which dif-
ferentiates the subject matter of synchronic linguistics from that
of diachronic linguistics. Certainly the original model involving
only the two temporal axes is inadequate, whilst the analogy with
chess, which forms the basis for Saussure's preferred model,
limps badly at a number of points.
We shall return to some of these issues in Chapter 7. But we
must now turn to Saussure's detailed account of the way in which
a language constitutes a very special kind of system, one of pure
values.
 LANGUAGE AS A SYSTEM
OF SIGNS
/ / / : Identities, system, and relations

Synchronic linguistics, it will be recalled, is 'concerned with

the logical and psychological relations that bind together co-
existing terms and form a system in the collective minds of
speakers' (CLG> 140, 99). Such a system is, Saussure claims,
one of pure values. But why must linguistic terms belong to
such a system? What are the relations which relate them to
each other? And why is a system of terms which are related
to each other by these relations one of pure values? Saus-
sure's answer to the first two questions is the principal sub-
ject matter of this chapter; Chapter 6 is devoted to discussion
of the third. Hence in these two chapters we shall trace the
way in which Saussure's answers to these questions lead to
the conclusion that *a linguistic system is a series of differ-
ences of sound combined with a series of differences of
ideas' the combination of which 'engenders a system of val-
ues'; moreover, 'this system serves as the effective link be-
tween the phonic and the psychological elements within each
sign' (CLG, 166, 120). But before we consider the detailed ar-
gument, I shall begin by raising what is at first sight a rather
curious question: What precisely are the terms of the system
in question?

88
 L A N G U A G E AS A SYSTEM OF S I G N S III 89

5.1. Concrete and abstract entities

The question seems curious, because in the light of the previous
discussion the answer appears to be obvious: They are linguistic
signs. But granted that that is so, why is it important from Saus-
sure's point of view to insist that 'the signs that make up language
are not abstractions but real objects (see p. 15); signs and their
relations are what linguistics studies; they are the concrete entities
of our science' (CLG, 144, 102)? In other words, what is the point
of insisting that the entities we are dealing with are real and
concrete?
Part of the answer is that, considered in isolation, signifiers
and signifieds are not linguistic entities: *A succession of sounds
is linguistic only if it supports an idea The same is true of
the signified as soon as it is separated from its signifier' (CLG,
144, 103). So from the point of view of linguistics, signifiers and
signifieds considered in isolation are abstractions, which is not
to say, of course, that they are abstract entities.
But the roots of Saussure's insistence that the entities in ques-
tion are concrete go deeper than this. The problems that he
grapples with derive from the second of his primordial princi-
ples, namely the Linear Nature of the Signifier (3.3). This prin-
ciple, it will be recalled, maintains that the elements of auditory
signifiers 'are presented in succession; they form a chain' (CLG,
103, 70). In other words, the principle tells us that if such a
signifier as [onehundredandone] is complex, so that there is a
question of what its constituent signifiers are, we can rule out
such a priori possibilities as that its constituent signifiers can co-
occur or be discontinuous.1 According to the principle, however
a complex sign is to be segmented, the resulting units must be
distinct both from what precedes them and from what succeeds
them in the chain. Hence Saussure's claim that 'the linguistic
entity is not accurately defined until it is delimited, i.e. separated
from everything that surrounds it on the phonic chain. These
delimited entities or units stand in opposition to each other in
the mechanism of language' (CLG, 145, 103). But how is this
delimiting to be done? The discussion of nu earlier (2.1) made
it clear that any attempt to delimit a linguistic unit in purely
physical terms would lack a rationale. For there would be in-
definitely many ways of segmenting any sequence in which nu
go SAUSSURE
occurred. But is it not possible that, whatever criterion we use,
there is more than one way of carrying out the analysis into
segments? In other words, is it not possible that there is a radical
indeterminacy in the procedure envisaged (7.2)?
Saussure apparently thought not, provided that we appeal to
meanings: 'Considered by itself, it [the signifier] is only a line,
a continuous ribbon along which the ear perceives no self-
sufficient and clear-cut division; to divide the chain we must call
in meanings' (ibid.). So, for instance, a reason for accepting
[hundred] as a signifier would be that it corresponds to a sig-
nified, whereas [andone] does not. It follows that the kind of
unit of interest to linguistics 'has no special phonic character,
and the only definition that we can give it is this: it is a slice of
sound which to the exclusion of everything that precedes and follows it
in the spoken chain is the signifier of a certain concept' (CLG, 146,
104). Moreover, Saussure goes on to describe a method of lin-
guistic analysis which commends itself in the light of this
discussion:
One who knows a language singles out its units by a very simple
method - in theory, at any rate. His method consists of using
speaking as the source material of language and picturing it as
two parallel chains, one of concepts (A) and the other of sound-
images (B).
In an accurate delimitation, the division along the chain of
sound-images (a, b, c) will correspond to the division along the
chain of concepts (ar, b\ c')\

a' b' c'

Take French sizlaprd. Can we cut the chain after / and make sizl
a unit? No, we need only consider the concepts to see that the
division is wrong. (CLG, 146, 104)
In fact, Saussure goes on to argue, there are only two possible
divisions, both of which are determined by the meaning ex-
pressed by the words: si-z-la-prd and si-z-l-aprd - respectively, si
je la prends ('if I take it') and sije Vapprends ('if I learn it').
One obvious consequence of this proposal is that concrete
 L A N G U A G E AS A S Y S T E M OF S I G N S III 91

units, whatever they are, cannot be identified with words. Some

units will be only parts of words - 'painful', for instance, can be
segmented into two elements, 'pain' and 'ful', each of which
clearly has meaning — whilst others will be longer than a word
- for instance, idioms, such as 'kick the bucket'.
But if a concrete unit is a slice of sound which signifies a certain
concept, what is an abstract unit and what makes it abstract? The
examples Saussure gives include word order (e.g., 'pain-ful' is
possible, but not '*ful-pain'), case and case endings, and parts
of speech. Since, for instance, a concrete unit may be a noun
but cannot itself be identified with the category noun, no part of
the speech chain can be identified with the category. Hence, the
latter is not concrete and so has to be treated as abstract, as do
case endings and cases, for similar reasons.
However, Saussure seems to have had more in mind than this
when he called units abstract, namely a doubt whether so-called
abstract entities are psychologically real. Their study is, he says,
difficult, 'because we never know exactly whether or not the
awareness of speakers goes as far as the analysis of the gram-
marian. But the important thing is that abstract entities are always
based, in the last analysis, on concrete entities' (CLG, 190, 138). In
other words, another major difference between concrete and
abstract entities is that whereas the former are psychologically
real, it is possible that the latter are not. Hence the possibility
that grammarians employ categories which are quite unfamiliar
to native speakers.2
This difficulty apart, it seems that Saussure employs two dif-
ferent criteria to determine whether an entity is concrete. Ac-
cording to the first, as we have just seen, a concrete unity is
psychologically real. But according to the second it must be de-
limitable in the speech chain in the way in which la is in sizlaprd
(SM, 210); and it is this criterion to which the method for singling
out units that we described appeals. But though the method is
'very simple in theory' (CLG, 147, 105), there are, Saussure points
out, difficulties in applying it in practice. For instance, though
it is tempting to treat words as concrete units, this would not be
correct: 'To be convinced; we need only think of French cheval
'horse' and its plural form chevaux. People readily say they are
two forms of the same word; but considered as wholes, they are
certainly two distinct things with respect to both meaning and
sound' (CLG, 147, 105).
92 SAUSSURE
However, if people do see two forms of the same word when
there are in fact two distinct concrete units, then arguably the
two criteria pull in different directions. And whether or not this
is so, one can see the practical difficulty of applying Saussure's
method by asking which if any of the following are segmentable
into two signifying constituents: [blackberry ]/[blueberry]/[straw-
berry]/[cranberry]/[bilberry]... As described, the method seems
to be of little help in cases like this in which, confronted with
the question whether or not [straw] is a significant constituent
of [strawberry], one's intuitions are feeble.3 Moreover, cases in
which there is a conflict of intuitions - e.g., between two people
one of whom wishes to segment la in sizlaprd into /, which signifies
the pronominal third person, and a, which signifies the feminine
gender, and the other who treats it as an unanalysed whole —
clearly raise pertinent questions (Harris 1987, 109). How do we
decide who is right?
Saussure's answer to this question is clear. After describing
the procedure for analysing sizlaprd, he writes:

To verify the result of the procedure and be assured that we are

dealing with a unit, we must be able in comparing a series of
sentences in which the same unit occurs to separate the unit from
the rest of the context and find in each instance that meaning
justifies that delimitation. Take two French phrases laforsdiiva (la
force du vent 'the force of the wind'), and abudufors (a bout du
force 'exhausted'; literally 'at the end of one's force'). In each
phrase the same concept coincides with the same phonic slice
fors; thus it is certainly a linguistic unit. But in ilmdfprsaparle (il
me force a parler 'he forces me to talk')/prs has an entirely different
meaning: it is therefore another unit. (CLG, 146, 104)

This brings to the fore the question whether a given tranche of

sound has the same meaning in different contexts; it is only if
it has that it can be treated as a signifier, and the union of it and
its associated meaning treated as a sign. In other words, it raises
for Saussure the crucial question of synchronic identity; that is,
of how we tell, for example, whether two occurrences of fors are
constituents of the same sign. As we shall see, not only do Saus-
sure's answers to the difficulties we have been discussing depend
on his account of synchronic identity, but in order to give that
account he has to explain the precise way in which signs form
a system.
 LANGUAGE AS A SYSTEM OF SIGNS III 93

5.2. Why signs form a system

Saussure's answer to the question of synchronic identity has two
parts. The first part consists of an argument to the effect that
utter chaos would result if the effects of the principle of the
Arbitrariness of the Sign were not restricted or diminished in
some way. The second, and more important, tries to show that
the identification of a particular signifier or signified depends
on its belonging to a system, because, paradoxically, it is not any
positive characteristic that it has which makes it what it is; what
is important, rather, are the ways in which it differs from the
other elements of the system. We must now consider each of
these answers in turn.

5.2.1. Relative arbitrariness

Though a linguistic system is based on the principle of the Ar-
bitrariness of the Sign, Saussure argues that it is a principle
'which would lead to the worst sort of complication if applied
without restriction' (CLG, 182, 133). However, it is not in fact
applied without restriction, since the mind 'contrives to intro-
duce a principle of order and regularity into certain parts of the
mass of signs' which is 'a partial correction of a system that is by
nature chaotic' (ibid.). As a result, some signs, e.g., 'twenty-one',
are said to be only relatively arbitrary, in contrast to those, like
'one', which are completely arbitrary.
Two characteristics mark a sign that is relatively arbitrary. It
is complex and has a model. That is, it is constructed on the
basis of a productive pattern, and because that is so, it necessarily
involves a system. Thus, 'twenty-one', 'thirty-one', etc. are only
relatively arbitrary, since they are constructed on a model,
whereas 'three', since it has no model, is radically arbitrary. How-
ever, I argued earlier that to call the phenomenon relative ar-
bitrariness is misleading, since the fact that a sign is constructed
systematically does not make it less arbitrary than one that is not
(3.2.1). In a metric system, 5op is just as much an arbitrary unit
of money as lp, even though it is part of a system. So system is
not, as Saussure suggests, a degree of arbitrariness.
Indeed, Saussure's claim that if applied without restriction the
principle of the Arbitrariness of the Sign would lead to the worst
sort of complication, i.e., chaos, is something of an understate-
94 SAUSSURE
ment if it suggests that we can conceive of a language that is not
systematic. For any language that has a potential infinity of sen-
tences must, if it is to be learnable, be systematic in a very precise
sense; that is, it must be what Lyons calls productive. By this
term he means 'that property of the language-system which en-
ables native speakers to construct and understand an indefinitely
large number of utterances, including utterances that they have
never previously encountered' (1977, 76). The crucial impor-
tance of this property was, of course, first stressed by Chomsky
(1957; 1965). So the task of linguistics is not simply to devise a
description of a corpus of sentences, however large, but to de-
scribe the system knowledge of which enables native speakers to
produce and understand indefinitely many sentences of their
language.
The importance of system can, therefore, hardly be doubted;
indeed, it is clearly of such crucial importance that it is difficult
to see why, having recognised its importance, Saussure should
not have seen the need to modify radically his claim that the
principles of the Arbitrariness of the Sign and of the Linearity
of the Signifier are the fundamental principles of a semiotic
enquiry. For system is surely an independent principle, and not
one that follows from the principles of arbitrariness and linearity,
as he claims.
But if a language must be systematic, is it essential that every
one of its signs belong to that system? A simple sign can, of
course, occur as part of a complex sign, as does 'one' in 'twenty-
one', and in that sense form part of a system. However, the
argument so far has not shown either that, their occurrence in
sentences apart, all simple signs do so occur or, more important,
that to identify or define such a sign one has to know that it forms
part of a system and what role it plays in that system. On the
contrary, it might be argued that it is only if simple signs are
independently identifiable and interpretable that the complex
signs of which they form part are interpretable. If simple signs
are not identifiable and interpretable independently of the other
elements of the system to which they belong, then it is not obvious
that they can be identified at all, for it is difficult to see how in
that case anyone could master the system, since there is no point
of entry to it.
So the fact that languages necessarily involve a system explains
why many signs, viz. complex ones, can be identified and inter-
 LANGUAGE AS A SYSTEM OF SIGNS III 95
preted only by someone who has grasped the system. It also
explains why, because they are systematically related to complex
signs, there is a sense in which many, if not all, simple signs
belong to a system. However, it does not show that simple signs
too can be identified and interpreted only by someone who has
grasped the system; and indeed the very idea that this is so runs
counter to deeply held intuitions about the relation of the simple
to the complex shared by philosophers as diverse as Locke, Leib-
niz, and Russell.
Nevertheless, Saussure certainly wants to argue that simple
signs can only be so identified and interpreted. So, as it stands,
the argument about the importance of system would seem to be
insufficient for his purpose. What is needed at this point is an
argument which shows that the identification of any sign,
whether simple or complex, depends on the identification of a
system of which it forms part.

5.2.2. Identities and differences

The crucial point that Saussure wants to make at this stage4 is
most clearly illustrated by his answer to the question of syn-
chronic identity, the importance of which I stressed at the end
of 5.1: In virtue of what can two occurrences of the same word
be said to be the same? Suppose, for instance, that someone
utters the word 'gentlemen' twice in the course of a speech, but
with very different intonations on the two occasions. Why, in
spite of all the differences there are between the separate speech
events, do we recognise two occurrences of the same word? It
can hardly be because of an absolute identity of the phonic sub-
stance on the two occasions, given the difference in emphasis
and intonation; nor need there have been an 'absolute identity
between one Gentlemen! and the next from the semantic view-
point either' (CLG, 151, 108). What makes them the same, in
spite of the differences, is the fact that there is one set of ways
in which each of them differs from all the other words of the
language. As for the admitted differences there are between
them, these do not entail that they are not instances of the same
word, any more than the myriad differences there are between
two oak trees do not entail that they do not belong to the same
species. In other words, some differences entail membership of
96 SAUSSURE
different kinds, and others do not. If all did, then there would
be no such thing as kind membership.
At this point Saussure develops an analogy:

For instance, we speak of the identity of two '8:25 p.m. Geneva-

to-Paris' trains that leave at twenty-four hour intervals. We feel
that it is the same train every day, yet everything - the locomotive,
coaches, personnel - is probably different what makes the ex-
press is its hour of departure, its route, and in general every
circumstance that sets it apart from other trains. (CLG, 151, 108)

Saussure hastens to add that this does not make the train an
abstract entity, since it must have a material realisation. Though
there is no particular group of carriages and so forth that it must
consist of, nevertheless it must consist of some. Thus the identity
of the successive trains cannot be accounted for in terms of the
material entities of which they are composed, for these might
be completely different. What makes the two trains the same, in
spite of their material differences, is the fact that each is related
in the same way to a time of departure, a route, a set of con-
nections with other trains, etc.
Another analogy that he develops to illustrate his point in-
volves yet another comparison with chess:

Take a knight, for instance. By itself is it an element in the game?

Certainly not, for by its material make-up - outside its square and
the other conditions of the game - it means nothing to the player;
it becomes a real, concrete element only when endowed with value
and wedded to it. Suppose that the piece happens to be destroyed
or lost during a game. Can it be replaced by an equivalent piece?
Certainly. {CLG, 153, 110)

Moreover, Saussure argues, the replacement need not be made

of the same sort of stuff as the original, or even be the same
shape and size. Presumably what matters, then, is the role it
plays in the game, and the specification of this requires one to
relate it to the other pieces and to the board.
If we call the kind of identity condition relevant in the case
of the 8.25 p.m. Geneva-Paris express and the knight 'relational'
- to distinguish them from the sorts of condition relevant to the
question whether a piece of stuff is the same, which might be
called 'material' - then Saussure's view is, of course, that the
 LANGUAGE AS A SYSTEM OF SIGNS III 97
conditions relevant to the question whether two occurrences of
the word 'gentlemen' are the same are relational, not material:
'Each time I say the word Gentlemen! I renew its substance; each
utterance is a new phonic act and a new psychological act. The
bond between the two uses of the same word depends neither
on material identity nor on sameness of meaning' (CLG, 152,
109).
Though this is plausible enough, it is important to note
that the position that Saussure wants to maintain is not sim-
ply that the identity conditions are relational. For he main-
tains that they are relational in a very strong sense, namely
that the relations in question are purely differential; by
which he means that the conditions in question mention only
ways in which the entities in question differ from each other,
without characterising them positively in any way - 'in lan-
guage there are only differences without positive terms' (CLG,
166, 120). If we call identity conditions of this sort 'purely
differential', then it is clear that whilst sets of purely differ-
ential conditions must be relational, it is not clear that every
set of relational conditions is purely differential. For instance,
it is conceivable that the 8.25 p.m. Geneva-Paris express was
related to another train because it connected with it at Lyon,
and that this was one of its identity conditions. But it is hard
to see what could be meant by saying that all we have here is
a difference between it and other trains which involves no
positive terms. Or, to take a different example, the fact that
the knight in chess moves as it does would seem to be a posi-
tive fact about it, and one moreover which is not obviously
relational at all.
So at this stage we must consider what kinds of relations Saus-
sure thought could hold between linguistic entities, and what it
was about them which entails that the identity conditions of
linguistics entities are not merely relational, but purely dif-
ferential.

5.3. Linguistic relations

The key discussion of linguistic relations5 is hard to follow be-
cause of the fact that Saussure often speaks vaguely of linguistic
terms, or indeed of words, and hardly ever makes use of his
canonical terminology of'signs', 'signifiers', and 'signifieds'. Inev-
g8 SAUSSURE
itably, that leaves considerable latitude of interpretation, but it
seems clear that what he wishes to maintain is this:

(i) That there are two generic kinds of relation that signifiers
and signifieds can enter into, syntagmatic and associative
(ii) That the specific kinds of syntagmatic and associative rela-
tions which relate signifiers to signifiers are different from
those which relate signifieds to signifieds
(iii) That the identity of a particular signifier or signified is purely
differential
(iv) That a system of signs results from the pairing of a system
of signifiers with one of signifieds; moreover, signs as such
have a value and are not purely differential entities
I shall discuss the first two points in the remainder of this
section and the other two in the next chapter.

5.3.1. Syntagmatic and associative relations

There are, Saussure argues, two different kinds of relation that
linguistic terms can enter into. The first, which he calls 'syntag-
matic', is based on the the linear nature of language and holds
between terms which are 'chained together', so that syntagmatic
relations are sequential. One example he gives is that of the
relation between 're-' and 'read' in the syntagm 're-read', so called
because it is a combination of elements which are syntagmatically
related. Saussure says that syntagmatic relations hold between
an item and everything that precedes or follows it, so that we
can presumably say that in 'God is good', which he cites as a
syntagm, 'is' stands in a syntagmatic relation to 'God' and to
'good'. In other words, 'God good' is a context in which 'is'
can occur.
Whilst the details of his argument are perhaps not very clear,
it is evident that Saussure thinks that both the importance of
and the need for syntagmatic relations follow from the principle
of the Linearity of the Signifier. Hence the importance which
he attaches to the principle, on which he claims depends the
whole mechanism of language (3.3). So for him it is not simply
a matter of brute fact that syntagmatic relations are important;
this is a fact that follows from a fundamental semiological
principle.
But is the fact that 're-' and 'read' are related syntagmatically
a fact about langue or one about parole? It would seem that
 L A N G U A G E AS A SYSTEM OF S I G N S III 99
Saussure must give the first answer; otherwise, he would not be
in a position to say in what way langue is a system. But, somewhat
disconcertingly, he writes as though it is because of the nature
of discourse that such relations exist. Such a relation is, he says,
l
in praesentia. It is based on two or more terms that occur in an
effective series' (CLG, 171, 123). All this suggests that syntag-
matic relations are, after all, facts of parole.6 We shall see how
Saussure attempted to resolve this tension later; but for the time
being it is important to note that unless syntagmatic relations
hold between elements of langue it is difficult to see that any
account can be given in Saussurean terms of the way in which
the latter form a system.
The second relation, which Saussure calls 'associative', exists
in virtue of similarities of form and meaning which hold between
a given item and other items to which it is not syntagmatically
related. For example, 'teaching' might be associated with the
group of terms {teaching, (to) teach, (we) teach...}, in which
there is something in common between the idea expressed and
its signifier in each case; or it might be associated with the series
{walking, singing, dancing...}, because they all contain a verb
stem and the signifier '-ing', which signifies the same thing in
each case; or it might be associated with the series {training,
lecturing, tutoring, educating...}, in virtue of similarities of
meaning; or it might be associated with the series {teasel, teatime,
teapoy...}, simply because of the perceived similarity of the
speech sounds realising the initial syllable of the signifier.7
Clearly, though these are all relations of the same generic type
— that is, they are all associative relations — they are different,
depending on the basis of the comparison (cf. (ii) in 5.3). For
instance, the second group is structure-dependent and depends
on similarities between signifiers and signifieds at a certain point
in that structure, whereas the third group is not structure-
dependent and concerns only contrasts between signifieds, i.e.,
ones which describe ways in which a teacher may impart infor-
mation to a pupil.
Unlike syntagmatic relations, associative ones are 'outside dis-
course' and 'are not supported by linearity' (CLG, 171,123). Thus
there is no difficulty in principle in maintaining that these re-
lations relate elements of langue and therefore that 'their seat is
in the brain; they are part of the inner storehouse that makes
up the language of each speaker. [Moreover] the associative re-
lation unites terms in absentia in a potential mnemonic series'
1OO SAUSSURE

(ibid.). So if there is a worry about associative relations, it is not

because it is hard to see how in principle they relate items be-
longing to langue. The worry is rather one about the vagueness
of the idea of relations that rest simply on the fact that groups
of terms have 'something in common'. Will there not be far too
many such relations? And won't many of those of the fourth
kind delineated — i.e., ones resting on perceived similarities be-
tween speech events that realise signifiers, such as that between
'teach' and 'teatime' - invariably be of no linguistic interest?
However, as we shall see, Saussure's implicit answers to the
two main problems mentioned so far (the question whether syn-
tagmatic relations belong to langue or parole and the worry about
the vagueness of the expression 'associative relation') are con-
nected.

5.3.2. Language, syntagmatic relations, and vagueness

The question whether syntagmatic relations pertain to langue or
parole is one which reveals considerable tension in Saussure's
thought. For though he cites a number of sentences as examples
of syntagms ('God is good', 'If the weather is nice we'll go out'
[CLG, 170, 123]) and in his lectures argues that the notion of a
syntagm can be applied to sentences as well as to complex words
(Engler 2, 283), he also maintains that it follows from the langue/
parole distinction that the sentence is a unit of parole.
There are, nevertheless, many more or less explicit hints in
the students' notes of a solution to the problem. There is, to
begin with, a very clear statement: 'This question of the order
of sub-units in the word recalls exactly that of the place of
words in the sentence' (Engler 2, 278B).8 So Saussure had one
powerful reason for treating the syntagms 're-do' and 'God is
good' in the same way. Either both belong to langue or both
to parole.
Second, as Saussure goes on to say, syntagmatic and associative
relations are interdependent: ' . . . they mutually condition each
other. In fact, spatial co-ordinations help to create associative
co-ordinations, which are in turn necessary for analysis of the
parts of the syntagm' (CLG, 177, 128). For example, consider
the verb 'un-do' (French de-faire). The syntagmatic relations be-
tween its parts, can be represented thus:
 LANGUAGE AS A SYSTEM OF SIGNS III 101

un-do

Moreover, the associative series 'comprising units that have an

element in common with the syntagm' can be represented as
follows:

un-do —

un-glue do
\
un-pick re-do

un-dress etc.
/ \
etc.

Now, it is clear that the possibility of relating the two represen-

tations in this kind of way is not for Saussure a contingent matter:
To the extent that the other forms float around defaire [un-do]...
these words can be decomposed into subunits. This is just another
way of saying that they are syntagms. Defaire [un-do] could not be
analysed, for instance, if the other forms containing de- [un-] or
faire [do] disappeared from the language. It would be but a simple
unit, and its two parts could not be placed in opposition. (CLG,
178, 129)

In other words, a condition of the syntagmatic opposition be-

tween 'un-' and 'do' is that each of them belongs to an (a dif-
ferent) associative series. But it is equally true that a condition
of each of them belonging to an associative series is that they
should stand in syntagmatic opposition to each other.9
So it would seem that Saussure must treat both kinds of re-
lation in the same way; either both belong to langue or both to
parole. But since the latter option would leave him with no ac-
count at all of the way in which langue is systematic, the only
tenable one is the first. Moreover, as we saw, sentential syntagms
have to be treated in the same way as morphological ones. So
the conclusion to which Saussure is deeply committed is that
sentential syntagms belong to langue.
1O2 SAUSSURE
Why then did he hesitate to draw this conclusion? Part of the
difficulty is clearly connected with his attempt to draw a dis-
tinction between the concrete and abstract objects of linguistics.
Since the categories needed to describe sentential syntagms are
abstract, something would have to give at this point. Second,
drawing this conclusion might seem to diminish the role allotted
to parole in Saussure's theory, in that it would leave next to no
choice for a speaker, leading us to the conclusion that, to par-
aphrase Levi-Strauss, it is language which speaks through men
rather than men through language.
Even so, the text of CLG, together with its sources, which are
more detailed on this point, shows that Saussure did see how to
incorporate into his system the conclusion that sentential syn-
tagms belong to langue. Doing this calls for a distinction between
a type and its exemplification or specimen:

To language [langue] rather than to speaking [parole] belong the

syntagmatic types that are built upon regular forms When a
word like indecorable arises in speaking... its appearance supposes
a fixed type, and this type is in turn possible only through re-
membrance of a sufficient number of similar words belonging to
language (impardonable 'unpardonable', intolerable 'intolerable', in-
fatigable 'indefatigable', etc.). {CLG, 173, 125)

Because of his worries about abstractions, Saussure insists 'that

types exist only if language has registered a sufficient number
of specimens'. But, subject to this qualification, he is prepared
to countenance sentential types; for example, one corresponding
to the sentence Que vous dit-il? 'What does he say to you?' In this
case what one does, he says, is to treat an element within a type
as variable:

lui
Que
me
dit-il?
vous
nous (Engler 2, 294E)

This is typical of the knowledge we possess of what to vary within

a unity to make a difference. Moreover, the probable mecha-
nisms involved in speech involve both syntagmatic types and
associative groups: 'We speak uniquely by means of syntagms,
 LANGUAGE AS A SYSTEM OF SIGNS I I I 103
and the probable mechanism is that we have these types ofsyntagms
in the head, and that at the moment of employing them we bring into
play the associations' (Engler 2, 294B). Thus, if a Frenchman
utters Marchons! ('Let's walk') it is the oppositions between it and
the forms Marche! ('Thou walk') and Marchez! ('You walk') that
affect his choice:

In reality the idea evokes not a form but a whole latent system
that makes possible the oppositions necessary for the formation
of the sign. By itself the sign would have no signification. If there
were no forms like marche! marchez! against marchons!, certain
oppositions would disappear, and the value of marchons! would
be changed ipso facto. (CLG, 179, 130)

So Saussure's way of dealing with his worry about the abstract-

ness of the descriptions of sentential syntagms is to insist that
they are based on specimens, whilst his solution to the problem
of choice in parole is to allow the speaker freedom to choose
between opposed items in a fixed frame.
It has, of course, to be conceded that his notion of types was
largely embryonic, and that he gives no more than a sketch of
how items on different levels are to be related.10 Even so, Saus-
sure was clearly in a position to treat sentential syntagms in the
same way as he did complex words, and thus treat syntagmatic
relations in the same way as he treated associative ones, that is
as items belonging to langue.
Moreover, the interdependence of syntagmatic and associative
relations means that the idea of the latter is not so vague as the
original description of it — as a relation holding between linguistic
units having something in common — suggests it is. For there is
a formal constraint on items related in this way, namely that they
must stand in opposition to each other in a context; that is, they
must be substitutable for each other in a syntagm. So we can
ignore the fourth type of example Saussure cites, the one in
which all that two units have in common is a similarity between
the speech events which realise their signifiers.
Finally, it should by now be clear why Saussure wants to main-
tain that the identity conditions for every sign are relational. For
apart from its syntagmatic relations, together with the constel-
lation of associative relations that it enters into, the sign 'would
have no signification' (CLG, 179, 130). Thus it is only because
104 SAUSSURE
Marchons! is opposed to Marchez! and Marche! that it has the
value it has. But the same principle applies equally to more
complex syntagms, including sentences:

To frame the question que vous dit il? 'What does he say to you?'
the speaker varies one element of a latent syntactical pattern, e.g.
que te dit il? 'What does he say to thee? que nous dit il? 'What does
he say to us7 etc., until his choice is fixed on the pronoun vous.
In this process, which consists of eliminating mentally everything
that does not help to bring out the desired differentiation at the
desired point, associative groupings and syntagmatic patterns
both play a role. (CLG, 180, 130)

The principle that linguistic units have no significance apart

from their syntagmatic and associative relations, and so neces-
sarily belong to a system, is the central thesis of Saussurean
structuralism. For on it depends the determination of what Saus-
sure calls values: 'Whatever might be the order of relations, the
word is above all a member of a system. This is important for
the determination of value; but even before speaking of value,
it is necessary to assert that words appear as terms of a system'
(SM, 90; Engler 2, 251). But before we turn to Saussure's de-
scription of the relation between system and value, we should
note that whilst his argument that the identity conditions of
linguistic terms are relational seems compelling, it is not clear
that they are purely differential. We shall return to this issue
later (7.4).

5.4. Summary and prospect

In this chapter we have been concerned with why, in Saussure's
view, linguistic items must form a system, and the nature of the
relations which hold between the elements of that system. We
began by considering why Saussure was so concerned to insist
that the entities with which linguistics is concerned are concrete
and not abstract. This was not simply a reminder that to consider
signifiers and signifieds on their own is to treat them as abstrac-
tions, but the expression of a worry that so-called abstract en-
tities, e.g., cases and case endings, are not psychologically real.
If they are not, then the kinds of analysis that linguists have
done have no foundation in the 'nature of things'. Furthermore,
 LANGUAGE AS A SYSTEM OF SIGNS III 105
there is clearly a worry that since linguistic analysis involves seg-
mentation of the phonic chain, there may be indefinitely many
ways of doing this. But in that case which, if any, of the possible
analyses can claim to be the correct one?
Saussure's answer to this question is that it is the one which
correctly reconstructs the underlying system to which the con-
crete entities belong. But why must linguistic units belong to a
system? The first answer Saussure gives, namely that it is because
many signs, though arbitrary, are only relatively so, rightly draws
attention to the importance of system, even if the way in which
it does so is confused. However, the most that this argument can
establish, it seems, is that the identification of complex signs
involves relating them to a system. So a further argument is
needed to show that the identification of all signs depends on
so relating them.
This argument raises the central question of synchronic iden-
tity: In virtue of what can two occurrences of a word be said to
be the same? Saussure's answer to this question, in the course
of evoking a number of forceful analogies with the 8.25 p.m.
Geneva-Paris express and the knight in chess, maintains that
there are two reasons why we count two separate speech events
as instances of the same word, e.g., 'Gentlemen'. First, each of
them differs in the same way from instances of other words in
the language; second, the differences there are between them
are not of any systematic significance. Because this is so, their
identity conditions are not material, but relational. Moreover,
they are relational in a very special sense; namely, they involve
only differences without positive terms, and so are purely
differential.
Clearly, to know in what ways instances of the same word differ
from instances of other words in the language, one must know
in what kinds of linguistically significant ways words and other
linguistic units may be related to each other. There are, Saussure
argues, two such ways. First, they might be related as 're-' and
'do' are related in 're-do'; that is, syntagmatically. Or, second,
they might be related as 'walking' and 'singing' are related, in
virtue of certain similarities of form and meaning; that is, as-
sociatively. Indeed, Saussure argues that all linguistic units are
systematically related to other such units by both kinds of
relation.
But do such relations pertain to langue or to parole? Clearly,
106 SAUSSURE
only if they pertain to langue has Saussure succeeded in principle
in giving an account of the way in which langue is a system of
signs, yet he says that syntagmatic relations pertain to parole.
However, I have argued that by distinguishing syntagmatic types
and their instances he was in principle able to treat both types
of relation as ones pertaining to langue, and, moreover, that he
was therefore in a position to treat sentential syntagms in the
same way as morphological ones. Furthermore, he was clearly
committed to doing this, even though there were a number of
reasons why he hesitated to do it.
Thus the foundations of Saussure's account of the way in
which a langue is a complex system of entities are now laid. There
are, however, two unfinished pieces of business. First, though it
is clear why the identity conditions of the elements of such a
system are relational, it is not so far clear why they are purely
differential. Second, it remains to be seen what is the relation
between belonging to such a system and having a special kind
of value, namely a pure value. It is to these issues that we now
turn.
 6

LANGUAGE AS A SYSTEM
OF SIGNS
IV: Values, differences, and reality

We have now reached the climax of Saussure's argument, and

the chapter in which it is developed, Part 2, Chapter 4, is arguably
the most important in the book. However, not only does the
chapter raise a number of very difficult issues, such as the re-
lation of language to thought and reality, but it is also a more
than usually complicated compilation of sources, so it is at times
far from easy to follow. Moreover, as we shall see, Saussure's
argument is not helped by the development of two incompatible
metaphors to explain his key term Value'. The prominence given
to these much-quoted metaphors tends to overshadow his central
thesis that in a language items which contrast with each other
reciprocally define each other's value, even though this is a thesis
which arguably stands or falls independently of the validity of
the metaphors.
We shall begin by tracing Saussure's introduction of the term
Value' and considering the ways in which it differs from the
related terms 'sense' and 'signification'. Next we shall consider
Saussure's explanations of why the values in question are pure
values, that is ones which arise from the system itself, and are
in no way determined extrasystematically. Finally, we shall ex-
amine the account given of the connection between values and
differences before going on to discuss Saussure's paradox that
107
io8 SAUSSURE

Figure 6.1
'in language there are only differences without positive terms.
Whether we take the signified or the signifier, language has
neither ideas nor sounds that existed before the linguistic system,
but only conceptual and phonic differences that have issued
from the system' (CLG, 166, 120).

6.1. System and values

Values are first and foremost products of a system for Saussure:
' . . . even before speaking of value, it is necessary to state that words
present themselves as terms of a system' (SM, 90). T h e system in
question is, of course, the set of syntagmatic and associative re-
lations that hold between the concrete entities of a langue. But
precisely how does the term Value' relate to terms already in-
troduced: 'signified', 'concept', and especially 'signification'?
Saussure admits that 'value' appears to be synonymous with
'sense' or 'signification', but nevertheless insists that it is distinct.
Recalling the schematic representation of the sign (fig. 6.1), he
identifies its signification with the counterpart of the signifier,
which is, of course, a signified, adding that to think of a sign or
term in this way is to consider it as an isolated absolute whole.
But this is in his view a partial and misleading view. For the sign
or term can also be seen at the same time as the counterpart of
other signs or terms (fig. 6.2). The terms of which it is the
counterpart are, of course, the other terms of its langue to which
it is related syntagmatically or associatively; and it is these re-
lations which determine its value. Hence: 'Language is a system
of interdependent terms in which the value of each term results
solely from the simultaneous presence of the others' (CLG, 159,
114). Now, since values are necessarily the product of a system,
the term 'value' would seem to have quite different connotations
from the term 'signification', since the latter as used traditionally
has no necessary connection with a system. So the question arises
 LANGUAGE AS A SYSTEM OF SIGNS IV 109

'Signified ^ ^Signified

whether there is any work left for the latter term in Saussure's
theory, since if its use arises from adopting a partial point of
view it is hard to see how it could be anything other than
misleading.
However, to understand Saussure's answer to this question,
we must look at the two not obviously compatible metaphors to
which he appeals to explain what a value is. According to the
first, values are like units of exchange; according to the second,
they are the products of the interaction of two otherwise undif-
ferentiated substances, thought and sound.1

6.2. Values and exchange

Appealing to an economic model, Saussure argues that, even
outside language, values are composed of two elements: '(1) of
a dissimilar thing that can be exchanged for the thing of which the
value is to be determined; and (2) of similar things that can be
compared with the things of which the value is to be determined'
(CLG, 159, 115). For instance, to know the value of a five-franc
piece one has to know what one can buy with it, e.g., a loaf of
bread, a newspaper, etc., and how it compares with 'a similar
value of the same system, e.g. a one-franc piece, or with coins
of another system (a dollar, etc.)' (ibid.). Analogously, a word
can be exchanged for an idea or compared with other words.
However, just as one cannot know the value of a five-franc piece
by knowing only what it can be exchanged for - for if that is all
one knows, one does not know whether one has paid a lot or a
little for one's purchase - one cannot discover the signification
of a word only by considering that for which it can be exchanged;
one must also relate it to the other words with which it can be
compared and with which it contrasts.
Saussure goes on to give a number of examples of the way in
which the systematic contrasts between a word and other words
determine its value. For instance, the French mouton does not
no SAUSSURE
mean the same as the English 'sheep', even though both might
be used with the same signification. This is so because in English
there is another word, 'mutton', which is used when speaking of
the cooked animal, whereas in French the same word signifies
live animal and cooked meat. In other words, there is a contrast
in English between 'sheep' and 'mutton' which does not obtain
in French between mouton and some other word.2 This is an
illustration of the principle that in a language words which con-
trast with each other reciprocally define one another's value.
Hence the content of a word 'is really fixed only by the concur-
rence of everything that exists outside of it' (CLG, 160, 115).
The principle that values are reciprocally defined by the re-
lations that obtain between terms that contrast with each other
applies, Saussure maintains, not only to words but to linguistic
terms of all kinds. For instance:

The value of a French plural does not coincide with that of a

Sanskrit plural even though their signification is usually identical;
Sanskrit has three numbers instead of two (my eyes, my ears, my
arms, my legs, etc. are dual); it would be wrong to attribute the
same value to the plural in Sanskrit and in French; its value clearly
depends on what is outside and around it. (CLG, 161, 116)

So in general we can say that the content or value of any linguistic

term is fixed by its comparison class, that is the set of terms with
which it contrasts.
Amongst the evidence cited by Saussure that this is so is the
fact that words do not have exact equivalents in meaning in other
languages, which they would do if they stood for pre-existing
ideas. Because of this we do not even find the same temporal
distinctions made in other languages: 'Hebrew does not recog-
nize even the fundamental distinctions between past, present,
and future' (ibid.), whilst the crucial lexicalised distinction made
in Slav languages between perfective and imperfective verbs is
difficult for a Frenchman to understand, since it is not made in
French. Hence, Saussure argues, in all of these cases we are
dealing with values which arise from the system. Moreover, if
these values are taken to correspond to concepts, 'it is understood
that the concepts are purely differential and defined not by their
positive content, but negatively by their relations with the other
 LANGUAGE AS A SYSTEM OF SIGNS IV ill
terms of the system' (GLG, 162, 117). It follows that the schema
which 'symbolises signification' cannot be taken as primitive (see
fig. 6.1). This is so because the signified is not determinable in
isolation from all the other elements of the system, so that the
schema is 'only a way of expressing the existence of a certain
value circumscribed in the system by its opposition to other
terms' (SM, 91).
The upshot of this discussion is, therefore, that the term 'sig-
nification' should only be used with the clear recognition that it
is parasitic on the term 'value'. As such it is clearly dispensable,
though in fact Saussure does not dispense with it. Indeed, there
are occasions on which he uses it where it does not seem to be
parasitic on the term 'value', e.g. in the sentence 'Modern French
mouton can have the same signification as English sheep, but not
the same value' (CLG, 160, 115). In these cases, though, it is hard
to see how it could be a term belonging to langue; presumably
Saussure's thought is that sometimes when speaking a French-
man may clearly be speaking about sheep and not mutton, even
though the distinction is not marked lexically. But this, if true,
is a fact of parole.
Saussure's economic metaphor obviously raises the question
of what precisely is being exchanged and what compared. At
this point it is important to distinguish between a unit of currency
and the coins or notes which have the value of the unit but which
are not to be identified with it (Harris 1987, 121). Now, since
what is exchanged in return for goods are coins, and what be-
longs to a currency system are its units, then if the analogy is to
be exact Saussure should maintain that what is exchanged are
tokens of signifiers for signifieds, and that it is signifiers that are
compared with other signifiers. But taken in this way the analogy
is doubly useless from Saussure's point of view. For, waiving
difficulties about the idea that tokens of signifiers are literally
exchangeable for signifieds, exchanges would on this account
belong to parole, since the occurrence of a token of a signifier is
a fact of parole. Second, it is difficult to see how the content of
a signified could be determined only by relations between sig-
nifiers, which, according to the analogy, are the entities corre-
sponding to units of currency. That is presumably why Saussure
speaks non-canonically of comparisons between words or else
between signs. But the latter of these comparisons clearly
112 SAUSSURE
departs from the analogy, since in a system of currency what is
compared are the units, not both the units and what they can
be exchanged for.
Another respect in which the analogy is less than ideal from
Saussure's point of view is that the goods for which coins can be
exchanged can have values other than monetary ones, so that
monetary values may in part reflect some other system of values.
But he clearly does not want to say anything analogous about
signifieds, for to do so would of course undermine the strong
version of the principle of the Arbitrariness of the Sign to which
he is committed (3.2).
It could, however, be argued that, flawed though the economic
metaphor is, Saussure's argument at this point does not depend
essentially on it. In particular, it might be said that his claim that
in a language items which contrast with each other reciprocally
define each other's values is a development of his thesis that
linguistic items can be identified only in terms of the syntagmatic
and associative relations they stand in to other items of a langue,
and as such should be assessed on its merits. Moreover, an in-
dependent argument, the translation argument, is advanced in
support of it. This is plausible; but it will be convenient to defer
the question of what remains of Saussure's argument without
the appeal to the economic metaphor until we have considered
the second of the metaphors adduced to explain the idea of a
value (6.4).

6.3. Thought—sound
The second metaphor involves a contrast between form and
substance, in which a langue is likened to a form which differ-
entiates otherwise amorphous substances. Considered in abstrac-
tion from language, ideas are, Saussure argues, amorphous, so
that 'nothing is distinct before the appearance of language' (CLG,
155, 112). Further, before the appearance of language, sounds
do not 'yield predelimited entities' - a conclusion which is hardly
surprising in the light of the earlier discussion of the French
word nu (2.1.1). The linguistic fact therefore has to be pictured3
'as a series of contiguous subdivisions marked off on both the
indefinite plane ofjumbled ideas (A) and the equally vague plane
of sounds (B). [Figure 6.3] gives a rough idea of it.' (CLG, 156,
112). So the effect of language on thought is not to create a
 LANGUAGE AS A SYSTEM OF SIGNS IV

Figure 6.3

phonic means for the expression of pre-existing ideas, as the

nomenclaturist would have us believe (3.1). The fact is that it
serves

as a link between thought and sound, under conditions that of

necessity bring about the reciprocal delimitation of units Nei-
ther are thoughts given material form nor are sounds transformed
into mental entities; the somewhat mysterious fact is rather that
'thought—sound' implies division, and that language works out
its units while taking shape between two shapeless masses. (CLG,
156, 112)

This somewhat mysterious fact is illustrated by an analogy: A

change in the atmospheric pressure of the air in contact with a
sheet of water will produce a series of waves, and it is these
'which resemble the coupling of thought with phonic substance'
(ibid.). It follows that 'linguistics then works in the borderland
where the elements of sound and thought combine; their com-
bination produces a form, not a substance' (CLG, 157, 113). But not
only are the elements combined in thought-sound amorphous
considered in themselves, but the choice of'a given slice of sound
to name a given idea is completely arbitrary. If this were not
true, the notion of value would be compromised, for it would
include an externally imposed element. But actually values re-
main entirely relative' (ibid.). Indeed, though Saussure does not
say this, this last point suggests an argument for both the claim
that values are relative and the claim that there are no pre-
linguistic ideas. For if the strong version of the principle of the
Arbitrariness of the Sign is true, and the identity conditions of
114 SAUSSURE
linguistic entities are purely differential (5.3.2), then it is hard
to see how values could be other than relative to a langue. More-
over, the existence of pre-linguistic ideas would have to be sus-
pect, since their existence might categorially motivate signs that
express them (3.2).
In this respect the current analogy suits Saussure's overall
position better than the economic metaphor does, since the latter
allows for the existence of independently determinable values.
However, if the strong version of the principle of the Arbitrar-
iness of the Sign should at best be treated as not proven, as we
argued it should, then any argument which, like the present one,
appeals to it should be treated with considerable caution. More-
over, reflection on the diagram and its accompanying explanations
suggests that in most respects the current analogy is much more
obscure than the economic metaphor. For the claim that a lan-
guage is a form which differentiates otherwise shapeless masses
strongly suggests that signifiers are slices of sound. But this is
difficult to reconcile with Saussure's claim that they are in fact
acoustic images, i.e., representations of sounds (3.1.1). Indeed,
he says explicitly that 'it is impossible for a sound alone, a material
element, to belong to language (langue)' (CLG, 164, 118). Fur-
thermore, the illustration of how two amorphous masses can
combine to produce a form is not very helpful. For the waves
are produced in the water by the action of the air pressure; but
surely thought, on Saussure's account, has no analogous role to
play in relation to sound. In short, the analogy throws no light
on the way in which the combination of the two masses, thought
and sound, could produce a form, for on Saussure's account no
agent or agency is involved.
Given his general theory, the only possible agents could be
individuals whose speech acts sustain or modify states of their
language without, of course, in his view, intending to do so either
individually or collectively. In other words, the appeal at this
point has to be to the comparison of a language with a game of
chess - bearing in mind, of course, the one fundamental dis-
analogy that whereas in the latter case players intend to bring
about the states that their moves produce, speakers do not intend
to bring about the changes in their language which result from
their speech acts (4.2). Hence speakers play a role in changing,
and indeed sustaining, their language without intending to do
so. Thus modified, the chess analogy provides a model of the
 LANGUAGE AS A SYSTEM OF SIGNS IV 115
way in which values depend on language states, and in turn of
the way in which these are the product of individual actions,
though not ones designed to bring states about. That is presum-
ably why, in the course of discussing the thought—sound meta-
phor, Saussure hastens to add that the fact that the sign is
arbitrary also explains why language has to be viewed as a social
fact: 'The arbitrary nature of the sign explains why the social
fact alone can create a linguistic system. The community is nec-
essary if values that owe their existence solely to usage and gen-
eral acceptance are to be set up' (CLG, 157, 113).4 So the
metaphor of thought-sound as both the product of the inter-
action of two shapeless masses and a form which creates units
within them simply does not stand on its own feet. Without the
implicit appeal to the chess metaphor, no account of how the
two masses are shaped, modified, and changed is available.

6.3.1. Divisions within a continuum

Even so, it might be argued that Saussure is making an important
point which is independent of the fruitfulness of his metaphor.
That is that different languages do provide systematic but dif-
ferent ways of differentiating things which are not as a matter
of brute fact differentiated, or, at least, make different and in-
congruent distinctions within what is intuitively the same do-
main. A striking example of this phenomenon, often adduced,
is colour terms. In the colour spectrum there are no sharp dis-
continuities. Moreover, it is prima facie difficult to see how colours
could be natural similarity classes, since, for instance, a shade of
red close to the borderline between red and orange will be more
similar to some shades of orange than it is to some shades of
red. So the drawing of a boundary between adjacent colours
seems to be an arbitrary matter. It should, therefore, not be
surprising that

it is an established fact that the colour-terms of particular lan-

guages cannot always be brought into one-to-one correspondence
with one another: for example, the English word brown has no
equivalent in French (it would be translated as brun, marron, or
even jaune, according to the particular shade and the kind of
noun it qualifies); the Hindi word pild is translated into English
as yellow, orange, or even brown (although there are different words
n6 SAUSSURE
for other shades of 'brown'); there is no equivalent to blue in
Russian - the words goluboy and sinij (usually translated as 'light
blue' and 'dark blue' respectively) refer to what are in Russian
distinct colours, not to different shades of the same colour, as
their translation into English might suggest. (Lyons 1968, 56)

These, of course, are only a few examples, but it could be argued

that there are many more; 'adolescent', for instance, belongs to
a system of classification of the ages of man different from that
rehearsed by Jacques in As You Like It; and according to a famous
study there was a major change in the structure of the conceptual
field of knowledge and understanding in Middle High German.5
This study was the work of Trier, who made a distinction be-
tween lexical and conceptual fields, the role of the former being
to divide the latter into parts. According to Lehrer,

he found, for example, that around 1200, part of the intellectual

field consisted of three terms that patterned in the following way:

Wfsheit

Kunst List

Kunst referred approximately to courtly knowledge, including

social behavior, List was used for more technical skills or knowl-
edge, and Wisheit was a more general term covering the whole
field. By 1300 the intellectual field had changed. Wisheit came to
be used in a religious or mystical sense, and Kunst was used for
more mundane skill and knowledge, having lost its connotation
of courtly and social knowledge. Wissen, a new term in the intel-
lectual sphere, was used for art, and List moved out of the semantic
field, (Trier 1931, 1934)

Wisheit Kunst Wissen

ca. 1300
(Lehrer 1974, 15)

But though these examples do provide important evidence that

vocabularies are structured in systematic ways, it is not clear that
 LANGUAGE AS A SYSTEM OF SIGNS IV 117
the underlying idea which unifies them, that of a continuum
which is divisible in different ways, is either one that Saussure
had in mind or one that is completely generalisable. To begin
with, Saussure's comparison of pre-linguistic thought to a Vague
uncharted nebula', though speculative, does not suggest that the
world itself is a vague uncharted nebula, or that it consists of
continua. Nor does it seem sensible to think of the vague un-
charted nebula of pre-linguistic thought as something which it-
self gets divided up, since, for instance, it is not clear what
boundaries it has. The kinds of examples we have been dis-
cussing have been concerned with relatively limited domains:
colour, ages of man, and kinds of knowledge. Although the
isolation of these domains of course requires considerable so-
phistication, it is essential that they be delimited for purposes of
the analysis.6 But what are the boundaries of thought itself? On
Saussure's initial account, thought is more like a cloud or a gas
than like a field; i.e., it has no determinate boundaries. Moreover,
we must distinguish the kind of precision that arises in our
thought from making distinctions within or about its objects from
the kind that arises when we make distinctions within or about
thought itself.7
Second, since the colour spectrum is hardly an object of every-
day experience, the literal applicability of arguments which de-
pend on there being different ways of dividing it up is doubtful.
It is surely conceivable that people have lived in an environment
in which there are no instances of certain parts of the spectrum.
Moreover, the work of Berlin and Kay strongly suggests that the
differences among colour vocabularies are by no means com-
pletely arbitrary (Berlin & Kay 1970). Though it is true that
speakers from different languages do not agree about the
boundaries between such adjacent colours as green and yellow,
speakers of the same language do not always agree about them
either. However, considerable uniformity is to be found if we
concentrate on the focal points for each colour - that is, the most
typical or paradigmatic examples of it - rather than on its bound-
aries. On the basis of a study of twenty languages supplemented
by additional data, Berlin and Kay argued that there is a hier-
archy in the importance of colours:

They find that all languages have terms for black and white. If
there is a third term, it will be red. If a language has four terms,
ll8 SAUSSURE
the fourth will be either yellow or green. Languages with five
terms have both yellow and green. A word for blue is the sixth
to emerge, and a term for brown is the seventh. If a language
has eight or more colour words, it will have words for purple,
pink, orange, gray, or some combination. (Lehrer 1974, 153)

It is, moreover, suggested that the order in which colour terms

were added to languages corresponds to the hierarchy, as does
the order in which children learn them. The details of the hy-
pothesis have been questioned,8 but at the very least it draws
attention to the ways in which apparent diversities may at a
deeper level of explanation exhibit considerable uniformity.
Moreover, given the salience of differences in luminosity and
their importance perceptually, it is perhaps not surprising that
the difference between black and white should have been uni-
versally lexicalised, whilst the very widespread lexicalisation of
the difference between red and green might be explained by
the fact that it has a neurophysiological basis (Lyons 1977, 248).
Third, if the literal applicability of arguments which depend
on there being an arbitrarily divisible continuum is at best highly
questionable in the case of colour, there are many other areas
of experience in which it is hard to see how it could apply at all.
It is surely only in a metaphorical sense that we can in many
domains be said to be presented with a continuum at all; for
example, of extents of water, of masses of trees. And even then
the principles of classification involved in our vocabulary seem
to be far too complex for it to be possible to account for them
on the model in question. Consider for instance terms in English
for extents of water: 'brook', 'bay', 'cove', 'estuary', 'lake', 'ocean',
'pond', 'pool', 'puddle', 'reservoir', 'river', 'sea', 'stream', 'tribu-
tary'. The difference among 'river', 'stream', and 'brook' is one
of size, as is that between 'sea' and 'ocean'. However, the differ-
ence between 'sea' and 'lake' is not one of size, but of the nature
of the water (salt/fresh), for a sea, like a lake, can be land-locked
(Sea of Galilee). The difference between 'lake' and 'reservoir' is
one of the intended use of the water. Complex though this sys-
tem is, it does not mark all distinctions which other languages
think worth marking - for example, that between a river which
empties into the sea and one which does not (fleuvelriviere) - and
no doubt it marks distinctions which other languages do not.
The idea that these are distinctions within a single continuum
 LANGUAGE AS A SYSTEM OF SIGNS IV 119
which is in some sense neutral between languages just does not
seem fruitful.
However, Saussure's claim that a signified is to be defined
relationally in terms of the associative and syntagmatic relations
in which it stands to other signifieds has no difficulty in principle
in accounting for the kind of complexity that I have been dis-
cussing, since there seems to be no reason why there should not
be a great variety of ways in which a given signified can differ
from other signifieds with which it contrasts in virtue of standing
in an associative or syntagmatic relation to it. Moreover, not only
does Saussure claim that it is because signifieds have to be defined
relationally that linguistic terms which contrast with each other
reciprocally define each other's values, but also, as we saw, the
claim is independent of the economic metaphor as a plausible
development in its own right of the thesis that a linguistic item
can be identified only in terms of the syntagmatic and associative
relations that it has to other items within a langue (6.2). So, since
Saussure's metaphors have only muddied the waters and there
seems to be a way of developing his argument independently of
them, it is time to turn our back on them and consider this
independent argument.

6.4. Differences, oppositions, and pure values

We saw that Saussure argued that the identity conditions of a
linguistic item are relational (5.2) and that the relevant relations
are syntagmatic and associative ones which hold between it and
other items of a similar kind in a langue. Now, it would seem
that the link envisaged between possessing a value and belonging
to a langue is as follows:

(T) The value of a linguistic item is determined by the set of

syntagmatic and associative relations that it enters into with other
items in a langue.

If that is so, then it is clear both why Saussure is so insistent that

the values of an item arise from a system and why what he himself
says about values is itself so dependent on his claims about the
identity conditions of linguistic items.
At this point two closely connected questions arise. First, pre-
cisely what kinds of things can have a value, and how many
12O SAUSSURE
different kinds of value are there? Second, why did Saussure
think that values are purely negative, involving only differences
without any positive terms?

6.4.1. Kinds of value

The answer to the first question might seem obvious, given the
way in which the term Value' was introduced as cognate to the
terms 'sense' and 'signification'; namely, that it is signifiers that
have a value. Moreover, the value of a given signifier is its sig-
nified, and this we saw has to be defined negatively in terms of
its 'relations with the other terms of the system' (CLG, 162, 117).
However, if what makes something a value is the fact that it has
to be so defined, then it is important to note that Saussure main-
tains that signifiers as well as signifieds have to be defined in this
kind of way:

The conceptual side of value is made up solely of relations and

differences with respect to the other terms of the language, and
the same can be said of its material side. The important thing in
the word is not the sound alone but the phonic differences that
make it possible to distinguish this word from all others, for dif-
ferences carry signification. (CLG, 163, 117)

Indeed, this follows, he claims, from the principle of the Arbi-

trariness of the Sign. Since the relation between a signifier and
its signified is radically arbitrary, no signifier is any more ap-
propriate than any other to express a given signified, so that it
can never 'be based on anything except its non-coincidence with
the rest. Arbitrary and differential are two correlative qualities'
(ibid.).
But if signifiers have to be defined differentially, and accord-
ing to (T) being so defined makes something a value, then ought
not signifiers to count as values as much as signifieds? An initial
response to this question is no doubt that surely signifiers have
values but are not themselves values. But in that case (T) is
insufficient to explain why signifieds are values, for, as we have
just seen, signifiers too have to be identified negatively in terms
of their relations with other terms of the system. Indeed, ac-
cording to Saussure, not only do signifiers as well as signifieds
have to be so defined, but so do phonemes.9 These are, he says,
 LANGUAGE AS A SYSTEM OF SIGNS IV 121
'above all else opposing, relative, and negative entities' (CLG,
164, 119). This is so presumably because we can say that the
phoneme /p/ occurs initially in the English word /pat/ only be-
cause it there contrasts functionally with /b/, Id, HI, etc. The
contrast is functional in the sense that substitution of one of the
other phonemes for it would produce a token of another word,
for instance /bat/, /cat/, /fat/, etc. Similarly, /a/ is functional in
/pat/ because it contrasts there with Id, III, etc. But there are
languages in which the contrast utilised in English between /p/
and Ihl is not significant, so that in these 'a p-sound and a b-sound
... are in free variation, in the sense that the substitution of the
one for the other in the same phonetic environment preserves
the type-token identity of the resulting forms' (Lyons 1977, 233).
In other words, whether a sound is functionally significant in
a language depends on what contrasts and oppositions are
counted as significant by the phonological system of the lan-
guage. It is this point that Saussure is making when he writes
that 1 can even pronounce the French r like German ch in Bach,
dock, etc., but in German I could not use r instead of ch, for
German gives recognition to both elements and must keep them
apart' (CLG, 165, 119). Thus we must conclude that signifiers
and signifieds are both values, but ones of different kinds, be-
cause the kinds of syntagmatic and associative relations that hold
between the former are different from those that hold between
the latter. Moreover, it is important to note that as well as being
a value, a signifier can be said to have a value, namely the signified
with which it is conventionally correlated.
So we must distinguish two claims:
(i) A linguistic entity is a value if it is constituted by the set of
syntagmatic and associative relations in which it stands to
other entities of the same kind, and
(ii) A linguistic entity has a value if it is correlated with a signified
But what then does it mean to say that values are purely negative?

6.4.2. Purely negative values

One thing that could be meant can be illustrated by reference
to phonemes, the important point being that for a phoneme to
play its role it is necessary and sufficient that it be distinct from
the other phonemes in the system. As Jakobson says,'... it would
122 SAUSSURE
be perfectly correct for us to say that taken in isolation the
[French] nasal a phoneme is nothing, because its sole value in
French is its non-identity with all the other phonemes of the
French language' (1978, 64). In other words, the French nasal
a phoneme is simply different from all the other phonemes of
the language, and lacks any positive content of its own, since
there is no answer to the question 'What does the French nasal
a phoneme signify?'
So presumably what Saussure meant by calling phonemes neg-
ative entities was that they lack any positive content, and that to
play their linguistic role all they need is to be different from the
other phonemes in the same system. Hence he would have
agreed with Jakobson's claim:

The linguistic value of the nasal a phoneme in French, and in

general of any phoneme in any language whatever, is only its power
to distinguish the word containing this phoneme from any words which,
similar in all other respects, contain some other phoneme. Thus sang is
distinguished from son, sein, ca> sceau, sou, si, su, etc. (1978, 61)

Thus we can say that though phonemes are values, since they
are determined by the reciprocal oppositions in which they stand
to other phonemes, they do not have values, since they signify
nothing.
But, as Jakobson urges, the characteristic of phonemes of
being purely negative, of lacking content, as well as of being
relative and opposing entities, is not a characteristic of other
linguistic entities, as Saussure claims. For instance, Saussure ar-
gues that letters are, like phonemes, purely negative and dif-
ferential:

(1) The signs used in writing are arbitrary; there is no connection,

for example, between the letter t and the sound that it designates.
(2) The value of letters is purely negative and differential. The
same person can write t, for instance, in different ways:

The only requirement is that the sign for t not be confused in his
script with the signs used for Z, df etc.
 LANGUAGE AS A SYSTEM OF SIGNS IV 123
(3) Values in writing function only through reciprocal opposition
within afixedsystem that consists of a set number of letters. This
third characteristic, though not identical to the second, is closely
related to it, for both depend on the first. (CLG, 165, 119)

Revealingly, Saussure argues that being purely negative and dif-

ferential, the second characteristic, is closely related to the third,
the existence of reciprocal oppositions within a fixed system.
However, not only does he not tell us what the nature of this
relationship is, but in the case in question the third characteristic
is present but the second is not. For the letter t not only is a
value, but it also has one, namely the phoneme l\l which it des-
ignates. As Jakobson says,

It goes without saying that the existence of a determinate system

of graphemes is a necessary prerequisite for the arrangement of
each letter. But the thing which is of primary significance here is
the specific, positive value of each grapheme. Of course the letter
beta must be distinguished from the letters alpha, gamma, delta,
etc., but the raison d'etre for the Greek grapheme beta is its des-
ignation of the phoneme b, and all other graphemes have a similar
task to perform. The graphic image functions as a signifier and
the phoneme as its signified. (1978, 65)

The crucial point is this: The characteristic of being purely neg-

ative and differential is not entailed by that of being constituted
by a set of reciprocal oppositions within a system. Thus what is
true of phonemes, namely that they have no positive content
and have a merely differentiating role, is not necessarily true of
signifiers and signifieds.
However, it might be urged that as a criticism of Saussure this
misses the point. For the positive content that a grapheme or
signifier has on this account arises only from the correlation that
exists between it and a signified. But the result of that correlation,
a sign, does not have a purely differential role. Moreover, this
is something Saussure would admit, for he claims that

although both the signified and the signifier are purely differ-
ential and negative when considered separately, their combination
is a positive fact; it is even the sole type of facts that language
has, for maintaining the parallelism between the two classes of
124 SAUSSURE
differences is the distinctive function of the linguistic institution.
(CLG, 166, 120)10

In other words, though it is not true that the only role that
signifiers have is a differential one (because of the correlations
that exist between them and signifieds), the fact remains that,
this correlation apart, they are constituted only by reciprocal
oppositions within a system. Hence, if we consider them only
from this point of view, they are purely negative and differential
entities, as indeed are signifieds. It is to the consideration of this
claim that we now turn.

6.4.3. Differences
Because of its intrinsic interest and philosophical implications,
we shall restrict our discussion to the case of signifieds. The claim
that signifieds are purely negative and differential entities has
arguably two different sources in Saussure's thought. One
source, connected with the metaphor of the amorphous pre-
linguistic masses of thought and sound, argues that the negative
and differential role of signifieds arises from the fact that they
occupy, or correspond to, an arbitrarily chosen part of a contin-
uum which has no positive features to distinguish it essentially
from other arbitrarily chosen parts. Culler, for instance, sees this
as an important source of Saussure's ideas:

We began by noting that there is no natural link between signifier

and signified, and then, trying to explain the arbitrary nature of
the linguistic sign, we saw that both signifier and signified were
arbitrary divisions or delimitations of a continuum (a sound spec-
trum on the one hand and a conceptual field on the other). (1976,
29)

However, we have seen that there are many reasons to doubt

the literal applicability of the idea that for every domain there
is a continuum which is arbitrarily divided by the terms belonging
to that domain (6.3.1). Moreover, there is something quite
un-Saussurean about the idea of the existence of language-
independent continua; that is, continua which are given, so
that their existence is independent of that of any particular
language.11
 LANGUAGE AS A SYSTEM OF SIGNS IV 125
A second source of Saussure's view that signifieds are purely
negative and differential entities rests on the thesis that signifieds
which contrast reciprocally define each other's value. Hence, the
value of a given signified is not given or language-independent.
What it is is determined by the language-specific relations in
which it stands to other items in the same language. Apart from
the fact that, unlike the first source, it does not presuppose the
existence of language-independent continua in each domain, this
argument has a further advantage. For the relations in question
could be very various and not just those required to partition a
given conceptual space or field.
Now, terms that differ from each other must contrast in some
way; A and B cannot just be different without being different
in some respect. However, the existence of a difference between
two terms does not always give rise intuitively to an opposition.
The signified 'chair', for instance, differs from the signified
'three', since the former but not the latter contains the compo-
nent (For sitting on), but intuitively they are not opposed terms.
In other words, difference is not sufficient for opposition.
However, if two signifieds belong to what is intuitively the
same domain, then difference does seem to give rise to oppo-
sition in the sense that 'X is an A' implies 'X is not a B\ and vice
versa: 'X is a By implies 'X is not an A'. For instance, 'X is a male'
implies 'X is not a female', and vice versa; 'X is hot' implies 'X is
not cold' and vice versa; and 'X is Tuesday' implies 'X is not
Sunday' and vice versa.12 In the first two of these cases the con-
trast is binary, but not in the third; for whilst male and female
exhaust the domain of gender, Tuesday and Sunday do not
exhaust that of days of the week. There is, moreover, a differ-
ence, pointed out by Lyons, between the first two cases of binary
opposition. For whereas 'X is not male' implies 'X is female', and
vice versa, 'X is not hot' does not imply 'X is cold' any more than
'X is not cold' implies 'X is hot'. This is no doubt connected to
the fact that the signifieds 'hot' and 'cold' stand at the end point
of a scale, in between which another pair of opposites, 'warm'/
'cool', is interposed (Lehrer 1974, 26). Furthermore, 'hot' and
'cold' are used to grade, so that 'This soup is hot implies that it is
hot relative to a certain implicit norm, either soup or perhaps
liquids served at a meal or even liquids in general. The norm is
different, however, in sentences like Paris is hot in the summer
(ibid.). Following Lyons, we shall call gradable opposites like
126 SAUSSURE

Differences

Oppositions

Non-binary Binary 'chair'/'three

Incompatibles Antonyms Complement

'Monday', 'hot'/'cold' 'man'/'woman

'Tuesday' . . .}

Figure 6.4

'hot'/'cold' antonyms, and reserve the term 'complementary' for

non-gradable binary opposites such as 'maleV'female' which mu-
tually exhaust some dimension of contrast, in this case gender.
Non-binary oppositions such as those involved in the set {'Mon-
day', 'Tuesday', ... , 'Sunday'} we shall cairincompatible' (Lyons
1977, ch. 9). Thus we have the distinctions shown in Figure 6.4.13
It is an important point that, though there would be no op-
positions without differences, not all differences give rise to op-
positions. Moreover, it is presumably only those differences
which do that are relevant to a signified's definition. In other
words, the set of signifieds the relations between which recip-
rocally define each other are not sets of signifieds which are
simply different, but ones which are opposed. This is important,
because since a given signified is different from every other
signified in indefinitely many respects, it is difficult to see how
it would be graspable if its characterisation depended only on
differences. We should be adrift in a sea of differences without
any principle of relevance. Moreover, oppositions, though they
involve differences, also involve similarities. For instance, 'man'
and 'woman' are opposed, since the latter contains the compo-
nent (Female) and the former does not. But for them to be
opposed in this way, both must contain the component (Animate),
 LANGUAGE AS A SYSTEM OF SIGNS IV 127
and there is no reason why they should not share all their other
components, e.g., (Human), (Adult). As Lyons points out,

we can say that X is married and Y is single, but in all other

respects similar. Moreover, we cannot predicate the word 'mar-
ried' and 'single' of X and Y, unless a certain number of other
words are predicable of X and Y. This holds for most, if not all,
lexical opposites. Oppositions are drawn along some dimension
of similarity. (1977, 286)

Presumably that is why only those differences which give rise to

oppositions are relevant to the determination of a signified's
content; for it is only if they do so give rise that the content of
the one can be said to be delimited by the other. But to reiterate,
this can only happen if they have something in common.
So far it is not difficult to see what Saussure might have had
in mind when he described signifieds as opposing and relative.
It is harder to see what he meant when he described them as
negative, claiming that 'in language there are only differences
without positive terms' (CLG, 166, 120). For, as we have seen, not
all differences are relevant to the determination of the content
of a signified; only those which give rise to oppositions are.
Moreover, the latter involve differences within some dimension
of similarity.
However, the binary thesis explains what might be meant by
saying that an aspect of a signified's content is not positively
characterisable. According to this thesis, a signified's content is
determined by a series of binary contrasts in which one term is
marked and the other unmarked; that is,14 'as Ihl may be said
to contain the phonological feature of voice, which /p/ lacks, so
(it might be said that) "man" and "boy"- contain the sense com-
ponent MALE, which "woman" and "girl" lack' (Lyons 1977, 322).
But obviously, whilst this explains what it means for a term to
be negatively characterised, it does so in a way which makes it
impossible for every term to be so characterised. For one member
of a binary opposition must be positively characterised for us to
be in a position to say that the other simply lacks a feature that
the first possesses. Moreover, there are some well-known diffi-
culties with this proposal. There are two different ways in which
a term can fail to contain the component (Male). In the first of
these, the lack implies the possession of some other characteristic,
128 SAUSSURE
i.e., (Female), so that instead of saying that 'woman' lacks the
component (Male) we might just as well have said that 'man'
lacks the component (Female). But in the second type of case,
the lack of (Male) implies nothing about the possession of any
other characteristic — from the fact that 'chair' does not contain
(Male) we cannot infer anything about what it does contain.
Commenting on the notation which uses '( + Male)' to denote
possession of the feature (Male), Lehrer writes:

The advantage of this notation is that it makes explicit the fact

that both features, e.g., [ +Animate] and [ — Animate] belong to
the same system. Unfortunately, the use of the — is inconsistent.
Sometimes it means that a feature is nonapplicable, and sometimes
it specifies a positive feature that contrasts with +, such as char-
acterising [Child] as [ — Parent]. If it is clear what positive feature
is meant when — is used in this way, there is probably no harm
in using this notation. (1974, 60)

Presumably, since it is clear what positive feature is attributed

in claiming that 'cow' has ( — Male) as a component, there may
be no objection to the use of the notation in this case. But 'bullock'
creates a problem, since to say that it lacks the component (Male)
is not to imply anything about its positive features. These prob-
lems apart, the treatment of gradable antonyms in a similar way
is, as Lehrer points out, extremely implausible. Apart from the
fact that it is not always plausible to characterise 'cold' as (— Hot),
part of the source of the difficulty is that they are opposed as
end points of a scale and so have to be related not only to each
other but to other terms in the scale.
It is of course true that there is no general rule that says which
of a dichotomous pair of features is dominant in a given case;
in the pair 'stallion'/'horse' it is ( + Male), in the pair 'lion'/'lioness'
it is ( + Female), whilst in the pair 'duke'/'duchess' neither would
seem to be. But whilst this might be taken to show that in the
language no feature is naturally dominant, it hardly shows that
particular signifieds cannot be positively characterised. Apart
from the ways I have already discussed in which this can arise,
there would seem to be a number of other such ways. For ex-
ample, a pig is an animal, a buttercup is a flower, etc., so that one
important part of the characterisation of the signifieds 'pig' and
'buttercup' is a specification of what kind or sort of thing pigs
 LANGUAGE AS A SYSTEM OF SIGNS IV 129
and buttercups are. But it is hard to see how this could be done
purely negatively; e.g., 'To be a buttercup is not to be any one
of the following: a rose, a peony, a delphinium, etc., etc/ For
apart from the fact that the idea of a closed list of kinds of flowers
is highly questionable, to be a buttercup is not just not to be
identical with any of the flowers mentioned in the list, for a dog
is that, but to be a flower that is not any of the other kinds
mentioned.
Another type of relation which would seem to need to be
positively characterised is that of a part to the whole. One could,
for instance, hardly understand what a collar is unless one under-
stood how it related to other parts of a shirt.

Even more convincing are sets of words like 'second', 'minute',

'hour', 'day', 'week', etc. The meaning of 'day', 'month' and 'year'
(and perhaps 'week') could be explained at least partly, without
mentioning any part—whole relationships that hold within the set.
... But it is in principle impossible to explain the meaning of
'second', 'minute' and 'hour' without specifying the part-whole
relationships holding within the set. (Lyons 1977, 314)

Thus a negative definition of the form 'A second is not a minute,

and not an hour and not a d a y . . . ' is not possible, because we
cannot say what a minute is without specifying its relation to a
second.
So if Saussure's claim that signifieds are purely negative en-
tities is true, we have completely failed to capture any sense in
which it is. Indeed, I have argued that only differences which
give rise to oppositions are relevant to the determination of a
signified's content, and that oppositions presuppose similarities.
Moreover, it seems that some of the ways in which signifieds call
for positive characterisation do not in any natural sense involve
oppositions. For instance, something can be both a buttercup
and aflower,so the signifieds 'buttercup' and 'flower' certainly are
not opposed to each other. And though, of course, an hour is
not a second, it contains seconds.
Thus, not only have we failed to make sense of the claim that
signifieds are purely negative and differential entities, but there
is good reason to think that there is no way of so doing. But if
that is correct, is Saussure's system not then in ruins?
To answer this question it is necessary to distinguish the prin-
130 SAUSSURE
ciple (T), that in a language items which contrast with each other
reciprocally define each other's value, from the principle (Ti),
that in a language there are only differences without positive
terms. We saw in our discussion of what Saussure has to say
about graphemes that he himself sees the two principles as being
closely connected (6.4.2). However, in at least one way of inter-
preting the phrase 'purely negative' there is no logical connection
between the two characteristics involved. So there is a strong
prima facie case for arguing that the truth of (T) stands or falls
quite independently of that of the much more dubious (Ti).
This is an important result, for the theory of structural linguistics
is strongly committed to (T); hence Lyons's claim that the 'de-
fining characteristic of modern "structural" linguistics is as fol-
lows: linguistic units have no validity independently of their
paradigmatic [ = associative] and syntagmatic relations with other
units' (1968, 75). But it is not similarly committed to (Ti); in-
deed, the way in which structuralist semantics has developed has
provided the basis for a reasoned critique of the principle. On
the other hand, structuralist and post-structuralist philosophers
often cite (Ti) as one of Saussure's major insights. In so doing
they have tried to build on one of the most opaque parts of
Saussure's theory, and one, moreover, which the actual practice
of structural linguistics does not presuppose. So the success of
that practice cannot, if I am right, be used as an argument for
the claim that in language there are no positive terms.

6.5. Summary
In this chapter we have been tracing the ways in which the
threads of Saussure's complex argument come together. In par-
ticular we have been concerned first with the connection he
makes between the notion of value and that of system, and sec-
ond with his claim that values are relative, opposing, and negative
entities, so that in languages there are only differences without
positive terms.
Since values are products of a system, it is important to note
that the theoretical implications of the term are radically differ-
ent from that of the term 'signification' which it supplants, since
they are holistic (3.1.1); we cannot first identify a value and then
describe the system to which it belongs, since 'words present them-
selves as terms of a system' (SM, 90). To illustrate the way in which
 LANGUAGE AS A SYSTEM OF SIGNS IV 131
values depend on a system, Saussure appeals to two metaphors.
In the economic metaphor he argues that, outside of language,
values have two features (6.2). There are dissimilar things for
which they can be exchanged, and similar things with which they
can be compared. For instance, a five-franc piece can be exchanged
for a loaf of bread and compared with a one-franc piece, a ten-
franc piece, etc. Analogously, a word can be exchanged for an
idea or compared with another word. And just as one can grasp
the value of a coin only if one knows what it can be exchanged
for and how it is related to other coins of different denomina-
tions, so one can grasp the value of a term only if one knows
both what it signifies and the relations that obtain between it and
terms to which it contrasts.
However, there are a number of respects in which the com-
parison breaks down. Apart from the obscurity of the notion
that words can be exchanged for ideas, the analogy says that
what would have to be exchanged are tokens of signifiers, so
chat the transactions would belong to parole and not langue. On
the other hand, comparisons would have to be between entities
corresponding to units of currency, which are signifiers, and not
their tokens; but it is difficult to see how such comparisons could
account for the values that signifiers have, that is the signifieds
with which they are associated.
In addition, the analogy has a very seriousflawfrom Saussure's
point of view. The goods for which coins are exchanged have
values other than monetary ones which, at least in part, deter-
mine their monetary values, so that it is difficult to see how the
latter could be a product of the system of currency alone. But
if they are not, then they cannot be pure values, and as such
radically arbitrary.
In this respect the second of Saussure's metaphors, that of the
interaction of two amorphous masses, wind and water, is much
better from his point of view (6.3). For the system of differences
arising from their interaction, the waves, has no existence in-
dependently of it. But there are other difficulties with this com-
parison. The diagram illustrating the analogy suggests that
signifiers should be construed as slices of sound - contrary, of
course, to Saussure's view that they are acoustic images (3.1.1).
And whilst the waves are a product of the action of the air on
the water, Saussure does not wish to allot an analogous role to
the thought mass. If language is a form (2.2), then we need some
132 SAUSSURE
account of the way in which 'something' which is neither thought
nor sound provides a principle for structuring both. And if at
this point to provide such an account we need to appeal to the
chess analogy, as I suggested, then the thought-sound metaphor
simply does not stand on its own feet.
As for the argument that, even so, the metaphor does draw
attention to an important truth - that languages provide prin-
ciples for making distinctions within phenomena which present
themselves as undifferentiated continua, e.g., the colour spec-
trum - I argued that it too is wanting (6.3.1). Apart from the
fact that the colour spectrum itself is hardly an object of everyday
experience, casting doubt on the literal applicability of argu-
ments that something which presents itself as undifferentiated
is then differentiated, the work of Berlin and Kay strongly sug-
gests that colour vocabularies are by no means completely ar-
bitrary. So, since Saussure's influential metaphors shed so little
light, I tried to describe the way in which his argument develops
independently of them, by tracing the relation between a value
and a system and the reasons for thinking that values are op-
posing, relative, and negative entities (6.4).
The connection between a value and the system to which it
belongs is, I argued, governed by the principle (T) that the value
of a linguistic item is determined by the set of syntagmatic and
associative relations that it enters into with other items in a langue.
Moreover, for Saussure not only are signifieds values, but so are
signifiers and phonemes (in the modern sense of that term),
though they are, of course, values of different kinds.
However, though signifiers, signifieds, and phonemes all are
values, the latter certainly do not have a value, since they are not
correlated with signifieds (6.4.1). Further, the distinction be-
tween being a value and having one is important for the eval-
uation of the claim that values are purely negative. One thing
that Saussure might have meant by this can be illustrated by
reference to phonemes. The important point is that for a given
phoneme to play its linguistic role — which, in Jakobson's words,
is 'to distinguish the word containing this phoneme from any words which,
similar in all other respects, contain some other phoneme' (1978, 62) —
it is necessary only that it be different from the other phonemes
in the language. It is not also necessary that it have a positive
content, i.e., that it should signify something. But if this is what
Saussure meant, it is true neither of letters nor of signifiers
 LANGUAGE AS A SYSTEM OF SIGNS IV 133
generally. They do not have a purely differentiating role, for as
well as being values they have values. Significantly, we noticed
that when talking about letters Saussure maintained that the
character of being purely negative is closely associated with that
of being relative and opposing, even though letters have the
latter characteristic but not the former, so that the association
cannot be logical. Clearly, phonemes cannot be taken as a model
of signifiers.
Finally, we considered the objection that Saussure could hardly
have denied that signifiers have values; after all, a langue is a
system for the correlation of signifiers and signifieds. That
granted, the objection continues, he maintains that, this corre-
lation apart, they are constituted only by reciprocal oppositions
within a system, so that if we consider them from this point of
view alone, they are indeed purely differential and negative, as
are signifieds.
However, we did not succeed in making sense of this claim in
the case of signifieds (6.4.3). Differences as such do not give rise
to oppositions; it is only in those cases in which signifieds (or
what they denote) belong to what is intuitively the same domain
that they do. Furthermore, though oppositions involve differ-
ences, they standardly involve similarities; as Lyons says, 'op-
positions are drawn along some dimension of similarity' (1977,
286). But why should the specification of the similarity not count
as a positive feature?
The binary thesis suggested an explanation of what it is for a
term to be negatively characterised - it is the unmarked term in
a binary opposition. But clearly this explanation requires other
terms to be positively characterised, other difficulties with the
proposal apart. Further, I argued that there are a number of
other important semantic relationships which call for a posi-
tive characterisation, e.g., kind membership and part—whole
relationships.
Thus not only did we fail to make sense of the thesis that
signifieds are purely differential and negative entities, but there
is good reason to think that this cannot be done. However, it
does not follow that Saussure's overall project shipwrecks. For
whilst the theory and practice of structural linguistics appeals to
the principle (T), it does not appeal to the principle (Ti), that
in language there are only differences without positive terms.
Moreover, (T) does not imply (Ti).
 SUCCESSES AND FAILURES

The preceding chapters have shown something of the scope and

ambition of Saussure's project. However incomplete it may have
been, it initiated a radically new perspective in the study of lan-
guage, a veritable Copernican revolution. This insisted that the
primary study of a language is a synchronic study of it as a system,
in which the main aim is to identify the units of the system and
the relations among them. Such units are not mere abstractions,
but concrete entities which are psychologically real (5.1), so that
there is in principle a determinate system to be reconstructed.
Moreover, their value is determined by the relations and op-
positions that obtain between them and other items of the system
rather than by their history or by other extraneous factors. Of
course, if this is the primary study, then it is essential that we
should be clear about what belongs to the system and what not;
hence the importance of the distinctions between langue and
parole and between synchrony and diachrony. And since the
system has to be described by describing the relations among its
elements, which are not identifiable independently of it, the im-
portance of Saussure's account of associative and syntagmatic
relations is also clear.
We saw that Saussure rejected the dominant organicist con-
ception of language (1.1). For him, the primary subject matter
 SUCCESSES AND FAILURES 135
of linguistics does not consist of linguistic forms developing in
accordance with principles applying to forms of that type, like
a species. It consists rather of signs which are radically arbitrary,
so that their significance is determined only by the conventions
of the historically constituted systems to which they belong. It
follows, he thought, that the synchronic study of language is
essentially one of social facts; that is, a study of the linguistic
conventions and relations operative at a given time in virtue of
which the signs of the system have the values that they do. But
since he was concerned that the fundamental units should be
concrete ones, he also thought that those conventions and re-
lations ought to be psychologically real. So if language is a study
of social facts, they are facts of social psychology, not ones which
exclude psychology from the study of language. Where nine-
teenth-century linguistics had 'divorced the study of language
from the study of mind' (Aarsleff 1967, 127), Saussure reinte-
grated the two.
As well as insisting that systems of signs form the primary
subject matter of linguistics, he also tried to develop a novel way
of studying them. This involved an appeal to a discipline which
did not exist when he wrote, but which had 'a right to existence,
a place staked out in advance' (CLG, 33, 16). That discipline is,
of course, semiology; and one of the most difficult parts of his
argument to evaluate is the attempt to derive fundamental
distinctions and conclusions from the semiological principles
proposed, beginning, of course, with the principle of the Arbi-
trariness of the Sign. But whatever the result of that evaluation,
it is clear that Saussure's conception of semiology had a major
impact on the development of structuralist thought in the human
sciences in general. For, as Culler says,

The importance of Saussure lies not simply in his contribution to

linguistics per se but in the fact that he made what might otherwise
have seemed a recondite and specialized discipline a major in-
tellectual presence and model for other disciplines of the 'human
sciences'. (1976, 53)

I shall have more to say about each of these themes in the

remaining sections. But before embarking on detailed discussion
it is worth speculating why Saussure did not think that his ideas
on general linguistics were in a state fit for publication. Perhaps
136 SAUSSURE
the main reason is that his arguments depended crucially on
there being a determinate system to reconstruct. Yet not only
does he not show how in detail to reconstruct such a system, but
his theory as it stands leaves it very unclear indeed that it is even
possible to do it in principle (7.2). Apart from his uncertainty
about what to class as concrete and what as abstract (5.1) - the
tensions in his system which lead him to treat sentences both as
units of parole and as syntagms and hence as units of langue
(5.3.2) — there is no reason to think that a determinate account
of a sign's associative relations can be given. Indeed, as Ducrot
has pointed out, there is every reason to think that it cannot:

Each sign is [according to Saussure] therefore related to all those

signs which delimit it, and which therefore constitute its paradigm.
But Saussure was not able to extract a criterion of classification
from his principle of negative limitation. The only phonic unities
which interested him were in effect signifiers, that is unities which
are already very complex, and which are as a result delimited by
a very large number of neighboring signifiers. One must therefore
accommodate within their paradign [ = set of terms to which they
are associatively related] a multitude of terms As a result, the
ordering of terms within a paradigm becomes an impossible task,
and furthermore it is not possible to classify them by relating
them to each other. (1968, 74)

But without such an account it is impossible, on Saussure's the-

ory, to determinately identify either the signs themselves or the
system to which they belong. We can identify a sign only by
identifying all its syntagmatic and associative relationships, and
the system to which it belongs is simply the set of all such rela-
tionships among all its signs.
Another reason why Saussure might have been unhappy with
his argument as it stood is that it leaves it unclear what the
linguistics of parole could consist of. But unless parole can be
systematically studied, then his theory leaves very little scope for
historical linguistics. If the mechanisms of change always involve
parole, then it is difficult to see what there is to say about them
unless there is a linguistics of parole. As the text stands, the
problem for someone who takes historical concerns seriously, as
I believe Saussure did (1.1), is not that they have been relegated
to a secondary position, but that it is quite unclear what a sys-
tematic study of them could be. The editors acknowledge that
 SUCCESSES AND FAILURES 137
the absence of a linguistics of parole is regrettable, but go on to
report: 'This study, which had been promised to the students of
the third course, would undoubtedly have had a place of honor;
why this promise could not be kept is too well known' (CLG, 10,
xxxi). So here is a lacuna which Saussure intended to fill, and
which urgently needs filling if a coherent methodology for dia-
chronic linguistics is to be developed (7.1).
A third reason why Saussure might have felt his argument
needed further development concerns its overall structure. If
he indeed intended to base it on semiological principles, then
surely a great deal more work needed to be done. For instance,
as we saw, he claims that the whole mechanism of language
depends on the principle of the Linearity of the Signifier, yet it
is quite unclear how it does if it does (3.3). And there are many
unresolved tensions implicit in the idea of a general science of
signs (7.4).

7.1. The fundamental dichotomies

The crucial importance of the fundamental dichotomies is clear,
for the distinction between langue and parole, on the one hand,
and between synchronic and diachronic linguistics, on the other,
is central to Saussure's claim that the primary object of study is
a synchronic one of a system of signs. I have already described
a number of unclarities in the way in which the first distinction
is drawn (2.5), but how serious are these for Saussure's system
as a whole?

J.I.I. Langue and parole

At least three unclarities have serious consequences. First, the
issue raised by Culler over the conflicting conceptions of langue
employed by Saussure brings into question his fundamental de-
cision to ground his theory on the notion of a sign. We saw that
the conflicting metaphors of langue as akin to Morse Code and to
an orchestral score correspond roughly to Hjelmslev's distinction
between schema and norm (2.4). But why does he have to choose
between the former, more abstract, conception, and the latter,
which includes the former? Part of the answer is that his theory
of the sign as a double entity makes it very difficult to adopt the
second conception. If one side of the double entity, the signified,
138 SAUSSURE
is allocated to the representation of meaning, then the other can
only represent the sign's Vehicle' at one level of abstraction; and
if there is only one level at which it can be represented, this has
to be the most abstract. If he really wished to take norms into
account and thus escape an excessively abstract conception of
language, Saussure needed to cease to rest his theory on the
conception of the sign as a double entity. Perhaps it does not
follow strictly from the conception of the sign as a double entity
that it cannot be represented on many more or less abstract
levels. But there seems to be little doubt that he did not explore
this possibility and, faithful to the conception of a sign as double
entity, maintained that a langue is a system for correlating two
distinct orders of differences, one phonic and one psychological
(CLG, 166, 120).
Second, the way in which Saussure defines parole means, we
saw, that it is at best an ill-defined rag-bag of what is left
when langue is subtracted from langage and at worst not de-
fined at all, since it is never clear precisely what does belong
to langage (2.1.3). If taken seriously, the implication that be-
cause of its very heterogeneity parole provides no subject mat-
ter worthy of study leads to a very narrow conception of the
scope of linguistics. Furthermore, as I said above, the failure
to articulate an account of parole also puts a stop to the devel-
opment of diachronic linguistics, since the mechanisms of
change always involve parole.
Third, the failure to show that langue is passive, in the
sense that the individual who acquires it by participating in
communicative interchanges plays no part in its acquisition,
threatens to undermine Saussure's principle of radical arbi-
trariness (2.1.2). If individuals are not wholly passive, then
the possibility that their language faculty plays some role in
the process cannot be ruled out. Just how active that role
might be is no doubt a matter for argument, but the failure
to exclude it threatens to wreck the exclusively semiological
approach. As things stand, Saussure has to concede the exis-
tence of at least one linguistic universal, the fact that all lan-
guages consist of systems of signs formally constituted by
their syntagmatic and associative relations. Unless this can be
shown to follow from fundamental semiological principles,
which I have argued it cannot, then Saussure has, on his own
account, to concede some contribution by the language faculty
 SUCCESSES AND FAILURES 139
(3.3, 7.4). Moreover, given the failure of his argument about
passivity, that contribution could be very great indeed, as
Chomsky has argued it in fact is (1965; 1966; 1969).

7.1.2. Diachronic and synchronic

The implications of the unclarities in the way in which the
distinction between synchronic and diachronic linguistics is
drawn for Saussure's system as a whole might seem less seri-
ous than those we have been looking at. His arguments that
the primary linguistic study is a synchronic study of langue
have come to be so widely accepted that it is genuinely diffi-
cult to imagine how anyone could have supposed anything
else to be the case. Yet this extraordinary achievement cannot
obscure the fact that Saussure did not succeed in delineating
the subject matter of diachronic linguistics; and, of course, if
its subject matter is not well defined, then, to the degree that
it is not, so must the subject matter of synchronic linguistics
be ill defined also.
We saw that the first model of diachronic linguistics, that of
two intersecting temporal axes, which implies that we are dealing
with one and the same thing studied from different points of
view, is clearly unsatisfactory. For synchronic linguistics is ac-
tually idiosynchronic, whilst items existing at the same time may
be studied diachronically (4.5). This strongly suggests that the
preferred analogy is with chess. But this analogy too, we saw,
seems to have serious drawbacks from Saussure's point of view.
In chess the time taken to effect transitions from one state to
another has no significance, so that the history of the game can
be represented as a series of non-overlapping states with no
intervals between them. But Saussure thinks that changes in
languages are not like that, since there may be periods of time
in which there is no stable state. In such a case, Saussure would
argue, the only possible study is a diachronic one. But apart
from the fact that it is difficult to see what this could be, given
his account of parole as it stands, the claim that no synchronic
description is possible in such cases entails that one cannot de-
scribe the langue internalised by the speakers of the language,
in spite of the fact that it is a fundamental principle of his theory
that there must be one. In short, at this point there is a totally
unresolved hiatus.
140 SAUSSURE

7.2. Language and determinacy

It is, of course, a fundamental principle of Saussure's system
that there is no parole without a corresponding langue, because
signs are constituted by their position in the latter. But is there
a determinate way of identifying the underlying system and of
relating a given sign to it? Ducrot, as we saw, argued that there
is not. The number of signs with which a given sign can be related
associatively and syntagmatically is indefinitely large, so that it
is not in principle possible to describe them all. But even if it
were, it is not clear that the resulting system would be deter-
minate, for reasons given by Quine.1
According to Quine's indeterminacy thesis,2 'manuals for
translating one language into another can be set up in diver-
gent ways, all compatible with the totality of speech dispo-
sitions, yet incompatible with one another' (i960, 27). In-
determinacy arises because there is no sense in which one
manual can be said to be right and the other wrong in the
areas in which they diverge. Moreover, the indeterminacy is
not simply a matter of underdetermination,3 for though the
competing translation manuals are indeed underdetermined
by the data, the distinctions they make are purely products
of the respective theories, and so in neither case reflect ante-
cedent distinctions:

... two translators might develop independent manuals of trans-

lation, both of them compatible with all dispositions to speech
behavior, and yet one translator would offer translations that the
other would reject. My position was that either manual could be
useful, but as to which was right and which wrong there was no
fact of the matter. (Quine 1977, 167)

Quine's thesis is vividly illustrated by considering the problems

of a linguist involved in radical translation, i.e., translation of a
language which has not been translated before. The point of
concentrating on this case is that it is one in which the linguist
has to adopt a naturalistic and behaviouristic account of mean-
ing, since there is no alternative.
In such a situation the linguist would, Quine argues, be able
to translate native observation sentences,4 such as 'Gagavi' stan-
dardly uttered when rabbits are present but not when other
 SUCCESSES AND FAILURES 141
animals are. He is able to translate this sentence because he is
able to observe the kinds of conditions in which native speakers
assent to/dissent from it, and can hence, subject to inductive
uncertainty, infer what their dispositions to assent to/dissent
from 'Gagavi' are.
Translation of this sort, Quine argues, 'can be objective
But the linguist's bold further step, in which he imposes his own
object-positing pattern without special warrant, is taken when
he equates the native expression or any part of it with the term
"rabbit" ' (1969, 2). This is a bold step indeed, because

given that a native sentence says that a so-and-so is present, and

given that the sentence is true when and only when a rabbit is
present, it by no means follows that the so-and-so are rabbits.
They might be all the various temporal segments of rabbits. They
might be all the integral or undetached parts of rabbits. (Ibid.)

It might seem that we could easily decide between these alter-

natives by identifying the native's sign for identity. But that,
Quine argues, is an illusion:

For if one workable overall system of [analytical] hypotheses pro-

vides for translating a given native expression into 'is the same
as', perhaps another equally workable but systematically different
system would translate that native expression rather into some-
thing like 'belongs with'. Then when in the native language we
try to ask 'Is this gagavi the same as that?' we could well be asking
'Does this gagavi belong with that?' (Ibid., 33)

Quine's own comment on this example is instructive. The ar-

gument depends on the fact that differing translations of a given
term can be compensated for by adjustments elsewhere in the
system. We could put the same point in Saussurean terms as
follows: Differing analyses of, or ways of segmenting, a complex
term are possible provided that (i) the differences between the
two analyses are compensated for elsewhere; and (ii) differing
sets of associative and syntagmatic relations are proposed in the
two cases. This is a conclusion which Saussure would not wel-
come, since it totally undermines the assumption that there is a
determinate system. Yet it is far from clear that he can escape
it, for, behaviourism apart, there are a number of striking sim-
 142 SAUSSURE
ilarities between Quine's approach to the study of language and
Saussure's.
First, Quine writes: 'With Dewey I hold that knowledge,
mind, and meaning are part of the same world that they
have to do with, and that they are to be studied in the same
empirical spirit that animates natural science. There is no
place for prior philosophy' (ibid., 26). Whilst Saussure cannot
be unreservedly associated with this naturalistic approach if it
is held to encompass behaviourism, there is no doubt that he
would have embraced many aspects of it. For example, he
criticised the Port Royal Grammar for its a priori approach;
poured scorn on the approach of Schleicher which 'led to
methods of reasoning which would have caused astonishment
in other sciences' (CLG, 17, 4); and insisted that linguistics is
concerned with concrete rather than abstract entities (5.1). In
other words, for him linguistics is a science with its own sub-
ject matter.
Second, Quine's objections to what he calls 'uncritical seman-
tics' bear a striking resemblance to Saussure's objections to the
idea that language is a nomenclature (3.1). For Quine, uncritical
semantics is 'the myth of a museum in which the exhibits are
meanings and the words are labels' (1969, 27). So, like Saussure,
Quine objects to the idea that words stand for predetermined
ideas. Indeed, in objecting to the myth of a museum Quine is
objecting not only to mentalism, but also to the identification of
meanings with objects of any kind "whether one assumes that
object to be an idea, a proposition, a physical body, or a Platonic
form' (Gibson 1982, 32).
Third, not only does Quine reject nomenclaturism but, again
like Saussure, insists that meanings are language-dependent:
'Meanings are, first and foremost, meanings of language. Lan-
guage is a social art which we all acquire on the evidence solely
of other people's overt behavior under publicly recognizable
circumstances' (1969, 26). Not surprisingly in view of this, Quine,
as well as Saussure, subscribes to a version of semantic holism
according to which the unit of meaning is not a sentence but a
language.
Thus it seems that the key steps on which Quine's conclu-
sion depends are ones that Saussure would accept. In par-
ticular:
 SUCCESSES AND FAILURES 143
(i) Complex terms need to be segmented if we are to analyse a
language
(ii) There is nothing extralinguistic which determines which of
the many possible ways of doing this is correct
(iii) Any analysis/segmentation uncovers a system
It is tempting to argue against the indeterminacy thesis that
any analysis has to reconstruct the phonology, morphology, syn-
tax, and semantics of a language as well as yield interpretations
for sentences in context that are pragmatically acceptable, and
that it is simply inconceivable that there should be radically dif-
ferent ways of doing this. But whether or not such an argument
is plausible, it is not available to Saussure. What plausibility the
argument has depends on its being true that the various sub-
systems of a language are subject to strong formal constraints.
But according to Saussure's theory the only formal constraint is
the weak one that the elements being analysed be related to each
other both associatively and syntagmatically. The strong version
of the principle of the Arbitrariness of the Sign to which Saussure
subscribes rules out the existence of any other constraints, formal
or otherwise. Indeed, it seems that in Saussure's case analyses
are even less constrained than in Quine's. For the latter, we at
least reach bedrock with the dispositions to assent/dissent of the
native speakers, whereas for Saussure there is no corresponding
constraint.
But if, given his overall assumptions, Saussure's method can-
not in principle yield a determinate description of the underlying
system, then he has clearly failed to achieve his fundamental aim
of identifying the object of linguistics. And, as Ducrot points out,
if the underlying system cannot be identified, then neither can
its components, given that they have to be defined relationally.

7.3. Language as social fact

As we shall see, the question of Saussure's commitment to the
kind of social realism espoused by Durkheim does not admit of
a simple answer, since there are important individualistic strands
in Saussure's thought. But before tackling this question, we had
better ask first just what was Saussure's relation to Durkheim.
Some have argued that it is not clear that there was one.
144 SAUSSURE

7.3.1. Saussure and Durkheim

We saw earlier that Mounin claims that the important sociological
dimension in Saussure's thought is one which has to be defined
in relation to Durkheim (1.1). However, Mounin argues, this is
not a straightforward matter; though they were more or less
exact contemporaries, they did not meet. The discussion of social
facts in Durkheim's Rules of Sociological Method makes only a
passing reference to language, and though in his article on so-
ciology and the social sciences in the Revue Philosophique in 1903
Durkheim proposed that language should be studied sociolog-
ically as a social institution, this hardly shows, Mounin concedes,
that Durkheim directly influenced Saussure. For though the latter
also invokes the metaphor of language as a social institution, so
too had Whitney, and Saussure had certainly read him.
In fact two questions need to be asked which are never very
clearly distinguished in Mounin's discussion: Did Saussure know
of Durkheim's ideas? Did those ideas positively shape his
thought?
Though Durkheim is never mentioned in CLG, there is ample
evidence that Saussure knew of Durkheim's work. He corre-
sponded regularly with Meillet, a former pupil of his who became
a professor at the College de France in 1906 and who had argued
that language fits Durkheim's definition, since it is independent
of particular users, but affects them causally. There is also the
evidence of Doroszewski, who wrote that he knew 'from a reliable
source that Saussure followed with deep interest the philosoph-
ical debate between Durkheim and Tarde' (1933, 90). And in
1957 he revealed his source: Louis Caille, one of Saussure's pu-
pils whose notes formed the basis for CLG.
Even without this evidence, it would be difficult to believe that
Saussure was not aware of Durkheim's ideas. The dispute be-
tween him and Tarde referred to above was carried on in jour-
nals for nearly a decade, culminating in a famous public debate
in 1903. It would be extraordinary if an original thinker in search
of analogies on which to base his account of language had over-
looked such a debate or was not aware of the Durkheimian
concept of a social fact (Sampson 1980, 48).
But if the answer to the question whether Saussure knew about
Durkheim's views is clear, the question of the latter's influence is
 SUCCESSES AND FAILURES 145
much more complex. In this connection the debate between
Durkheim and Tarde is instructive.
Durkheim was a social realist;5 that is, he believed that there
are social facts and that, moreover, explanations of them are not
generally reducible without remainder to facts about human
individuals. Such facts constitute the subject matter of sociology,
which is distinct from psychology precisely because it has its own
subject matter:

... one is forced to admit that these specific facts reside in the
society itself that produces them and not in its parts - namely its
members. In this sense therefore they lie outside the consciousness
of individuals as such, in the same way as the distinctive features
of life lie outside the chemical substances that make up a living
organism. (Durkheim 1982, 39)

Thus for him to treat suicide as a social fact is

to consider [it], not as an aggregate of individual acts, but as a

patterned phenomenon. I f . . . the suicides committed in a given
society during a given period of time are taken as a whole', we
find that 'this total is not simply a sum of independent units, a
collective total, but is itself a new fact sui generis, with its own unity,
individuality, and consequently its own nature'. The suicide rate
remains fairly stable in any given society from year to year
This stability affirms, according to Durkheim, that we are in the
presence of a social fact. For we can be certain that the individuals
who figure in the suicide rate in one year are not the same as
those who compose it in the next. There must be certain social
influences acting upon all these individuals. (Giddens 1978,43; the
quotation is from Durkheim 1952, 46)

But even if it is true that the explanation of social facts cannot

be reduced to psychological facts, a positive characterisation of
them is clearly needed.
The first chapter of Durkheim's methodological treatise, The
Rules of Sociological Method, is devoted to a discussion of this
question: It concludes with the following account:

A socialfact is any way of acting, whetherfixed or not, capable of exercising

over the individual an external constraint; or: which is general over the
146 SAUSSURE
whole of a given society whilst having an existence of its own, independent
of its individual manifestations. (1982, 59)

This seems to propose two characterisations of a social fact, one

in terms of the notions of externality and constraint, and one in
terms of generality with independence from any individual
instantiation.
By 'externality' Durkheim seems to have meant independ-
ence from any particular individual's will. Many norms or con-
ventions are, he argues, external in this sense. For though I
may have internalised them, 'it is not I who have prescribed
these duties; I have received them through education' (ibid.,
50). Unlike habits, which rule us from within, 'social beliefs
and practices act on us from the outside' (p. 44). Thus church
members find their beliefs and practices ready-made. As well
as being external, a social fact constrains individuals in the
sense that it imposes itself on them independently of their will.
It is true that in consenting to it they may not feel any sense of
constraint, but its existence manifests itself as soon as they try
to resist. So, according to the first of the two characterisations
of a social fact, it exists independently of the will of any of the
individuals it constrains.
The second characterisation of a social fact retains the idea
of externality, now defined as 'independence from individual
manifestations'. By 'generality' Durkheim simply refers to dif-
fusion of the social fact within a group, so that it has a wider
extension than facts of individual psychology, but a narrower
one than facts of biology. Clearly, this characterisation is not
equivalent to the first, since it omits reference to the notion
of constraint. One reason that Durkheim gives for adopting
it rather than the first is methodological; constraint can be
hard to identify (Durkheim 1982, 57). Another is, Lukes ar-
gues, that Durkheim wanted to count far too many different
things as instances of constraint for the concept to be useful
from his point of view (1973, 14). But though this is a plau-
sible criticism of Durkheim's use of the term, it is not for
that reason that he proposed the second characterisation. On
the contrary, the two characterisations are, according to him,
linked by the fact that 'if a mode of behavior existing outside
the consciousness of individuals becomes general, it can only
 SUCCESSES AND FAILURES 147
do so by bringing pressure on them' (1982, 57). In other
words, the second characterisation is de facto equivalent to the
first, because the requisite kind of generality with independ-
ence will obtain only when there is constraint, whether or not
we can identify the precise form the constraint takes. There
is, therefore, no need to mention constraint; when the condi-
tions mentioned in the second characterisation obtain, it will
be present. So, though not mentioned, it is important; and
when Durkheim wrote the preface to the second edition of
the Rules in 1901 he continued to emphasise the importance
of it: ' . . . this is what is most essential in the idea of social con-
straint. For all that it implies is that collective ways of acting and
thinking possess a reality existing outside individuals who, at
every moment, conform to them' (1982, 44). Though there un-
doubtedly are problems with Durkheim's employment of the
notion of constraint, it can hardly be omitted altogether from
his account of social facts.
Not surprisingly, publication of the Rules provoked a contro-
versy about the nature and existence of social facts, which was
pursued at times with great acrimony. One of Durkheim's lead-
ing critics, Tarde, was a methodological individualist and hence
believed that social facts have to be explained ultimately in terms
of facts about individuals. As Mill put it, 'Human beings in society
have no properties but those which are derived from, and may
be resolved into the laws of individual man' (1875, 469). Tarde
argued that imitation plays a major role in accounting for the
uniformity of social behavior, and so proposed a distinctively
individualistic account of social facts. But where does Saussure
stand in all of this? Is he a social realist or a methodological
individualist?6

J.3.2. The issue

To begin with, an important point is that methodological indi-
vidualism is, as its name suggests, a methodological and not an
ontological thesis, so that a proponent of it could perfectly well
allow that statements about social individuals are not logically
reducible to ones about human individuals (Danto 1973, 321;
MacDonald & Pettit 1981, io6ff). What the methodological in-
dividualist is committed to maintaining is rather:
148 SAUSSURE
(i) Social facts have to be explained causally in terms of facts
about individuals, and not vice versa
(ii) Though there may be explanations of social facts in terms
of other social facts, these are not ultimate (Danto 1973, 322)
So, if H is a predicate of human individuals and C one of social
individuals, then according to the methodological individualist
an ultimate explanation will contain laws of the form (la) but
no laws of the form (lb):

(1a) Vx Vy (Hx -+ Cy)

(lb) Vx Vy(Cy -* Hx)

In other words, the explanation will contain laws which account

for the properties of social individuals in terms of those of hu-
man individuals, but will not contain laws which account for the
properties of human individuals in terms of those of social in-
dividuals. By contrast, social realists will maintain that sometimes
ultimate explanations will contain laws of the form (ib) - though
they can of course concede that often ultimate explanations con-
tain only laws of the form (1a) - and so give themselves less to
prove than methodological individualists (D'Agostino 1986, 15).
It is not thus far clear what counts as a predicate of an indi-
vidual. However, it is clear both that this is a crucial issue and
that if a relatively relaxed view is taken of what counts, the victory
of methodological individualists is going to be too easy to make
their thesis interesting. They might, for instance, claim that an
explanation of the success of a certain party in an election in
terms of the high turnout of its members is an individualist one,
even though the predicate 'is a member of party x9 used in that
explanation remains unreduced. So it is necessary to place some
restriction on H; for instance, that it is a predicate which involves
no essential reference to a social individual or event and does
not ascribe to its bearer an irreducibly social attitude. Whether
this suggestion is the right way of making the restriction is a
moot point, but it is clear that what the restriction has to do is
to ensure that in ultimate explanations the only laws involved
are ones which, in Mill's phrase, are 'resolved into the laws of
individual man' (7.3.1). A restriction of this sort would seem to
be acceptable to methodological individualists. Watkins, for in-
stance, writes:
 SUCCESSES AND FAILURES 149
Society is a system of unobservable relationships between indi-
viduals whose interaction produces certain measurable sociolog-
ical phenomena. We can apprehend an unobservable social system
by reconstructing it theoretically from what is known of individual
dispositions, beliefs and relationships. (1973, 165)

Presumably the dispositions in question, whether ones possessed

by everyone, e.g., acquisitiveness, or ones possessed by only some
of us, e.g., bravery, are the subject matter of individual rather
than social psychology. Otherwise it is hard to see what force
there could be to the claim that something unobservable had
been reconstructed from something observable.

7.3.3. Saussure's individualism

Recalling that for Durkheim the characteristics of a social fact
are generality (but not universality) and external constraint,
meaning by the latter 'independence of any particular individ-
ual's will', then Saussure's characterisation of facts about langue
is strikingly reminiscent of Durkheim. Such facts are general in
the relevant sense and involve constraint, since according to
Saussure 'the individual does not have the power to change a
sign in any way once it has become established in the linguistic
community' (CLG, 101, 69). Indeed, we saw that in the course
of trying to explain why languages are so stable he goes even
further, claiming: 'No individual, even if he willed it, could mod-
ify in any way at all the choice that has been made; and what is
more, the community itself cannot control so much as a single
word; it is bound to the existing language' (CLG, 104, 71).
The reasons why this is so are, Saussure argues, connected
with the fact that language is a heritage from preceding gen-
erations (3.4). Language is checked by the weight of the collec-
tivity, since 'it blends with the life of society'; but it is checked
also by time: 'These two are inseparable. At every moment sol-
idarity with the past checks freedom of choice. We say man and
dog because our predecessors said man and dog' (CLG, 108, 74).
But if no one invents his or her language, and indeed passively
assimilates it, then it would seem to follow that there cannot be
an individualistic explanation of why, for instance, the plural of
'goose' in modern English is 'geese'. For facts about the language
I speak cannot be explained in terms of anything I have done,
150 SAUSSURE
since I am powerless to change the language I speak. It is some-
thing I did not invent but acquired passively by participating in
exchanges. But the same is true of my ancestors, and of their
ancestors, and so on. So it would seem that Saussure is committed
to claiming that
(1; No individualistic explanation of why 'geese' is the plural of
'goose' in modern English can be given
But what then are we to make of the actual explanation he
gives of the way in which this in fact came about? I described
this in detail earlier (4.2); the essential points are these. In mod-
ern English 'goose' is one of a number of nouns with plurals
marked by a specific vowel change. How these nouns came to
be this way involves three previous stages, the transitions between
which were brought about by phonetic changes that had nothing
to do with the plural per se. At the first stage the plural was
marked by the addition of a final i to the singular form, so that
the language contained the pair gos:gosi. Then, as a result of a
phonetic change which occurred whenever i followed a stressed
syllable, the contrasting pair became gos:gesi. The language now
made available in this sort of case, two ways of marking the
plural: the addition of a final i and the contrast between 0 and
e. Finally, at the third stage, as a result of the fall of the final i,
the language contained the pair gos:ges.
Now, if all the transitions involved were of this kind, then it
would be hard to resist Saussure's conclusion that the diachronic
facts responsible for changes in the way the plural is marked
were in no way directed at bringing them about; they were simply
the unintentional by-product of changes in the phonetic system.
Indeed, as we saw (4.2), one of the points that Saussure wished
to underline was that since the phonetic changes responsible for
the evolution of the various ways of marking the plural were not
designed to bring them about, an account of that evolution can-
not be teleological: 'Speakers did not wish to pass from one
system of relations to another' (CLG, 121, 84). But, of course,
the fact that the explanation is not teleological does not mean
that it cannot be individualistic; one has only to think of the
Darwinian theory of evolution, with which Saussure's account
shares some similarities, to see that. Perhaps, for instance, the
transition from Stage 2 to Stage 3 (i.e., from gos:gesi to gos:ges)
came about because some speakers omitted to pronounce the
 SUCCESSES AND FAILURES 151
final i and others followed suit; since there was no loss in the
expressive power of the language involved, there was no reason
why the variation should not become entrenched.
This is of course speculative. But it is consistent with Saussure's
claim that it is 'in speaking (parole) that the germ of all change
is to be found' (CLG, 138, 98). Moreover, it does draw attention
to the fact that an individualistic explanation need not be teleo-
logical; things might have changed as a result of things speakers
did, even though what they did was not designed to bring the
changes about. As Popper has pointed out, the route taken by
a path can be the result of innumerable individual acts, none of
which was done with the intention that the path should follow
the route it does.
Indeed, though it is speculative, the explanation is no worse
than one Saussure himself gives. This concerns the change in
the conjugation of the German copula in the sixteenth century,
which involved the replacement of was by war, so that the con-
jugation ich was, wir waren became ich war, wir waren ('I was', 'We
were'):

Some speakers, influenced by waren, created war through analogy;

this was a fact of speaking; the new form, repeated many times
and accepted by the community, became a fact of language. But
not all innovations of speaking (parole) have the same success, and
so long as they remain individual they may be ignored. (CLG,
138, 98)

So it would seem that, after all, Saussure ought to be willing to

concede that
(ii) An individualistic explanation why 'geese' is the plural of
'goose' can be given, though it is not, of course, teleological
This conclusion, on reflection, might riot seem all that surpris-
ing. It is certainly not self-evident that the characteristics of social
facts, generality and constraint, cannot be explained individu-
alistically.7 Consider Saussure's own question, raised when he
introduces the distinction between langue and parole, 'How does
the social crystallization of language come about?' (CLG, 29, 13)
— that is, how does it come to be the case that members of the
same linguistic community have broadly similar representations
of their langue} His own answer seems to be individualistic
152 SAUSSURE
(2.1.2). A group of people came to have similar representations
because they talk to each other, and as a result of their efforts
to interpret each other, 'impressions that are perceptibly the
same for all are made on the minds of speakers' (CLG, 29, 13).
Though short on detail, this account is individualistic. For Saus-
sure, langue is the product of face-to-face communicative inter-
changes between individuals, each of whom, as a result of these
interchanges, ends up with a similar representation of it. The
collective existence of these representations is then a social fact
because of the explanation of how they are acquired and the
role they play in further exchanges. For since this explanation
requires two or more individuals, it presupposes generality, one
of the characteristics of social facts mentioned by Durkheim.
Moreover, it guarantees externality, i.e., independence of any
particular individual's will, another of the characteristics he men-
tions. As for constraint, this arises for Saussure, as we saw, be-
cause of the way in which language is acquired, its widespread
diffusion, the effect of transmission over time, etc. (3.4.1). The
collective force of all these factors makes it very difficult for the
individual not to conform to the linguistic practices of other
members of the community.

7.3.4. Saussure as social theorist

The account above is an individualistic one. There was no point
at which it seemed to be necessary to invoke the idea of a col-
lective mind and its representations, unless indeed that is simply
a shorthand way of talking about representations in the minds
of individuals within the community, in explaining how the social
crystallisation of language came about. Nevertheless, it is not
clear that Saussure could accept that a thoroughgoing individ-
ualistic explanation of all aspects of language change and lan-
guage acquisition is possible. There are two reasons why this is
so.
First, consider the explanation Saussure gives of the change
of conjugation of the German copula in the sixteenth century,
in which ich was, wir waren became ich war, wir waren. As we saw,
Saussure's explanation is that some speakers created war by anal-
ogy with waren, and others followed suit, so that eventually the
new form displaced the old one. Now, for this to happen speak-
ers must have had beliefs about the form of the first person
 SUCCESSES AND FAILURES 153
plural in order to use it as a model for the first person singular.
However, it is difficult to see how such beliefs could be ascribed
to them without using predicates which refer to the German
language as it was then constituted. But since it is far from clear
that these predicates are in the relevant sense individualistic, i.e.,
ones which do not make an essential reference to a social indi-
vidual or ascribe to its bearer an irreducibly social attitude, it is
far from clear that this is an individualistic explanation. To be
sure, the only agents involved are human individuals; there is
no suggestion that social individuals or facts are themselves
agents. However, the human individuals would not have acted
as they did had they not had beliefs about social entities. So what
we have is not a thoroughgoing individualistic explanation.
It might, however, be objected that this is not an ultimate
explanation, and that reference to social entities would be elim-
inated in such an explanation. But on what grounds can it be
said not to be ultimate? It surely cannot be because it is incom-
plete in the sense that it does not explain how the German lan-
guage came to be in that state at that time. It is hard to see how
any explanation could be anything other than incomplete in this
sense. Presumably the point is that the very existence of any
laws, rules, or conventions invoked can be explained individu-
alistically, so that they are only intermediate generalisations. For
instance, it might be argued that it was only a convention that
the first person plural of the German copula was wir waren and
that the notion of a convention can be analysed individualisti-
cally, thanks to an analysis proposed by Lewis.
According to Lewis,

Conventions are regularities in action, or in action and belief,

which are arbitrary but perpetuate themselves because they serve
some sort of common interest. Past conformity breeds future
conformity because it gives one a reason to go on conforming;
but there is some alternative regularity which could have served
instead. (1975, 4)8

In other words, a convention is an arbitrary beneficial regularity.

For instance, driving on the left in England is a convention
because though the regularity is arbitrary - we could have driven
on the right - it is beneficial, since it is in everyone's interest to
drive on the same side of the road. And it does seem true that
154 SAUSSURE
past conformity provides a reason for future conformity. But
precisely how does it do so?
A key part of Lewis's answer to this question involves the
notion of mutual knowledge. There is mutual knowledge that q
when everyone knows that q, everyone knows that everyone
knows that q9 and so on. Now, according to Lewis not only does
everyone drive on the left, but it is mutually known that this is
so; moreover, it is mutually known both that the belief that
everyone else does it gives each individual a reason also to do
it, and that there is an alternative beneficial regularity. Thus,
because of mutual knowledge everyone can replicate everyone
else's reasoning, and so everyone can see that everyone else has
a reason to conform. So the mutual-knowledge 'condition en-
sures stability. If anyone tries to replicate another's reasoning,
perhaps including the replication of his own reasoning,... the
result will reinforce rather than subvert his expectation of con-
formity to R' (Lewis 1975, 6). In other words, every member of
P's beliefs about what everyone else in P is going to do reinforces
what would otherwise be a weak reason for personally conform-
ing to a regularity R.9
However, the appeal to mutual knowledge to explain the per-
petuation of the regularity makes it questionable whether this is
a genuinely individualistic explanation. For in the explanation
of how an individual coordinates his or her actions with those
of everyone else, it is necessary to attribute to everyone both a
belief about what everyone else believes and a belief about a
general preference for general conformity. In each case, pred-
icates involving an ineliminable reference to other members of
a group are involved.
The second reason for doubting whether Saussure would ac-
cept that a thoroughgoing individualist explanation of language
change can be given is that for him it is axiomatic that there is
no parole without langue. Hence any explanatory theory of parole
will presuppose, or draw on, a theory of langue; this is the whole
point of the metaphor of the platform on which the theoretical
study of language stands (2.1.3). Note that Saussure's account
does not attribute agency to langue — the only agents of language
change, according to him, are human individuals. His point is
that one cannot explain why individuals speak as they do without
ascribing to them beliefs about a langue.
In conclusion, three points. If the methodological individu-
 SUCCESSES AND FAILURES 155
alists' main point is that human individuals are agents but social
individuals are not, then Saussure has no quarrel with them as
far as I can see. However, if the methodological individualists
want also to insist that social facts can be explained without
attributing to human individuals attitudes towards social indi-
viduals, then Saussure cannot agree with them. But even if Saus-
sure is not a methodological individualist for this reason, that
does not mean that he is committed to denying that individu-
alistic explanations are ever appropriate in linguistics. As we saw,
a denial of methodological individualism commits one only to
the claim that sometimes individualistic explanations are inad-
equate.

7.4. Language, science, and signs

Saussure's conception of semiology raises two main questions.
First, to what extent was his own approach to linguistics se-
miological, and how successful was it? Second, what impact did
his conception of semiology have on the development of both
semiology itself and other disciplines? Clearly, the second of
these questions is a topic for a book in its own right; we shall
here concentrate on the first.10
There is no doubt that Saussure attached great importance to
a semiological approach to linguistics. Shortly after introducing
his conception of semiology as 'a science that studies the life of signs
within society' (CLG, 33, 16) he writes, 'I wish merely to call at-
tention to one thing: if I have succeeded in assigning linguistics
a place among the sciences, it is because I have related it to
semiology' (ibid.). One reason why he insisted on this relation
was no doubt to distance himself from the organicist conceptions
of the comparativists, which he rejected totally (1.1). Another
was that doing this underlined his social conception of language
by making it the subject matter of social psychology. But perhaps
the most important reason was that by comparing a language
with other semiological systems it would be possible to see what
were its essential and its non-essential features most clearly:

... to me the language problem is mainly semiological, and all

developments derive their significance from that important fact.
If we are to discover the true nature of language we must learn
what it has in common with all other semiological systems; lin-
156 SAUSSURE
guistic forces that seem very important at first glance (e.g., the
role of the vocal apparatus) will receive only secondary consid-
eration if they serve only to set language apart from the other
systems. (CLG, 34, 17)

If we study a language from the point of view of individual

psychology, or indeed of sociology, we are likely, Saussure
thought, to concentrate on more or less contingent features and
overlook the fact that it is a system of signs — its most important
characteristic. Only adoption of the semiological perspective will
enable us to see this clearly and to concentrate on the key ques-
tion. What are the essential features of such systems?
From this point of view, the way to begin the study of language
is to develop the theory of signs, to describe the principles that
hold for all signs and sign systems, and then to derive a descrip-
tion of linguistic sign systems from the theory and the principles.
It might be argued that the text of CLG should not have been
organised so that it began with the distinction between langue
and parole as though that were something ultimate. The text
should rather have begun with the theory and principles of se-
miology, from which the distinction between langue and parole
should have been derived.11
However, if this was Saussure's strategy, it is surely a ques-
tionable one. To begin with, it takes it for granted that different
sign systems have something interesting in common, in spite of
the fact that the examples given are very diverse, viz. 'a system
of writing, the alphabet of deaf-mutes, symbolic rites, polite for-
mulas, military signals, etc' {CLG, 33, 16). In fact, the first of
these is, according to Saussure, a secondary and parasitic system,
whilst the second, since it is alphabetic, has no morphology or
syntax of its own. Semaphore, one type of military signal, is like
the deaf alphabet in this respect, whilst naval signals using flags,
another type of military signal, are, according to Saussure's own
principles, unlike linguistic signifiers in that they are not con-
strained by the principle of linearity (3.3). Symbolic rites and
polite formulas, unlike words, do not enter into many interesting
syntagmatic relationships, so that they are not obviously part of
productive systems, i.e., ones with indefinitely many 'utterances'.
This may be so because they themselves are often more like
utterances than like words - e.g., greetings, farewells, congrat-
ulations, etc.12
 SUCCESSES AND FAILURES 157
Clearly, the question of the ways in which linguistic signs are
special is of crucial importance for Saussure's argument, yet it
surfaces only once in CLG, when the question of the extent to
which other signs are arbitrary is raised. No doubt Saussure is
right to insist that though polite formulas may be based on nat-
ural expressions they are nevertheless fixed by rule, for it is often
specified who ought to use them in whose presence and when.
However, the crucial comparison is not with rules of this sort
but with the rules that define the content of the formulas: Do the
formulas form systems within which they are related by ana-
logues of associative and syntagmatic relations? Prima facie, many
non-verbal signs do not form systems, but consist rather of ex-
aggerated ways of doing what one might be doing anyway. Take
what Goffman calls body gloss.

In our own middle-class society, for example, there is a stan-

dardized little Scarsdale smile, held over-long,... 'transfixed' so
that throughout the whole period of an offender's behaviour he
will be accorded a sign that no offence is being taken, that no
contest is involved, and that sympathy is present for whatever
alignment to the situation he chooses to take. (1972, 156)

In these sorts of cases conventionalisation seems to consist of

nothing more than a stylised and somewhat exaggerated way of
doing what one might be doing naturally. It is difficult to see
what the system is or could be.
Anyway, if Saussure's aim was to discover what is essential to
language by eliminating anything which is not common to all
sign systems, then he was clearly in very great danger of having
only a few very abstract characteristics left with which to char-
acterise it or, at worst, nothing at all. Moreover, such a strategy
is not obviously consistent with the idea that a language is an
autonomous social institution of a quite special kind, so it is not
surprising that there is a certain tension in Saussure's account
leading him to vacillate between two quite different positions.
According to the first, which I have been discussing, the strategy
is to begin with the laws of semiology and to derive the char-
acteristics of langue from them. But according to the second, the
way to proceed is to study langue as a semiological institution
and then to take stock and see what light this study throws on
semiology in general.
158 SAUSSURE
The attraction of the second position for Saussure is clear from
his insistence that language is the most important of the se-
miological systems, and that 'to determine the exact place of
semiology is the task of the psychologist. The task of the linguist
is to find out what makes language a special system within the
mass of semiological data' (CLG, 33, 16). Moreover, 'Linguists
have been going around in circles: language, better than any-
thing else, offers a basis for understanding the semiological prob-
lem; but language must, to put it correctly, be studied in itself
(CLG, 34, 16). Indeed, much of the time it would seem to be
this second modest strategy that is in fact followed. Saussure's
analysis of the sign, for instance, questionable though it may be
as a study of linguistic signs, would be even more so if it was
meant as one of signs in general: The conception of the signifier
as an acoustic image seems quite inappropriate for mathematics,
the deaf alphabet, semaphore, etc. So what his account of syn-
chronic linguistics starts with is, in effect, not an analysis of signs,
but one of linguistic signs.
I suggested earlier that, whatever his intentions, much of Saus-
sure's argument does not depend on his commitment to the first
of the two strategies distinguished above, which bases the ar-
gument on the theory of signs in general, but at most on the
more modest one, which is semiological to the extent that it
studies languages as a species of sign system but makes little or
no appeal to the theory of signs in general (1.1). This, of course,
does not rule out the possibility of using the results of that study
as a model for the study of other sign systems, which incidentally
would seem to be all that is necessary for the kind of application
to social anthropology of the work of structural linguistics en-
visaged by Levi-Strauss (1977, 31).
Certainly it would have to be conceded that if the first strategy
was Saussure's, its theoretical base was very restricted indeed.
Apart from the scepticism I expressed about whether all sign
systems have anything in common at all, and apart from the fact
that it is hardly plausible to take his analysis of the sign as one
of signs in general, his argument rests on just two primordial
principles. But of these the second, the Linear Nature of the
Signifier, on which the whole mechanism of language is said to
depend, is evidently not a general semiological principle appli-
cable to sign systems of every kind. For that matter, the first
principle, the Arbitrariness of the Sign, is not obviously so, either,
 SUCCESSES AND FAILURES 159
if the qualification 'radically' is taken seriously. However, it might
seem attractive at this point to make a definitional move and
restrict semiology to the study of sign systems whose signs are
arbitrary. And though Saussure does not in fact do this, he comes
quite close to doing so by arguing that semiology's 'main concern
will still be the whole group of systems grounded on the arbi-
trariness of the sign' (CLG, 100, 68). Even so, there are, he
acknowledges, degrees of arbitrariness; for instance, gestures
and symbols, such as the scales ofjustice, are not wholly arbitrary.
But presumably then the strong version of the principle needed
for language will not be appropriate in these cases (according
to the strong principle, linguistic signs are wholly, i.e., radically,
arbitrary). So even if the scope of semiology is restricted to sys-
tems which are to some degree arbitrary, it is not clear that there
is just one precisely specified principle of arbitrariness which
applies to them all.
And Barthes's claim that other semiological systems are, as it
were, parasitic on language has to be answered; this is a point
that he makes very forcibly:

It is true that objects, images and patterns of behaviour can sig-

nify, and do so on a large scale, but never autonomously; every
semiological system has its linguistic admixture. Where there is a
visual substance, for example, the meaning is confirmed by being
duplicated in a linguistic message... so that at least a part of the
iconic message is, in terms of structural relationship, either re-
dundant or taken up by the linguistic system. (Barthes 1967, 10)

Of course if this is so, then, as Barthes points out, other se-

miological systems cannot be independent of language. But if
they are not, there is no generalised semiology which could pro-
vide the theoretical starting point for a study of language.
Moreover, not only does the first strategy have a thin theo-
retical base, but its implementation leaves a lot to be desired. As
we saw, Saussure never establishes that signs are radically arbi-
trary in the sense of being totally unmotivated; indeed, many
possible forms of motivation are simply not discussed (3.2). Con-
nected, the failure to show that language acquisition is wholly
passive opens up the possibility that the language faculty itself
plays a substantial role in language learning (7.1). Third, the
attempt to derive from the principle of Linearity a theoretical
160 SAUSSURE
justification for the claim that the only linguistically relevant
relations are associative and syntagmatic ones not only fails, but
leaves one wondering how it ever could have been expected to
account for associative relations, which are now-linear by defi-
nition. Fourth, system, a key idea for the development of Saus-
sure's argument, is treated as a species of arbitrariness — relative
arbitrariness — whereas it is an independent idea (3.2.1). Fifth,
the conception of a sign that Saussure employs is not only not
a plausible candidate for one of signs in general, but assumes
uncritically that signs are double entities. So even assuming that
Saussure thought that his key distinctions could be derived from
his primordial principles together with his conception of a sign,
it is difficult to see that the derivation as such could be of much
interest.
No doubt Saussure's actual argument is much less tidy if, as
I have suggested, most of the time he followed the second modest
strategy. But his contribution of the reorientation of the study
of language was, for all that, epoch-making. It marked a decisive
reorientation in the history of linguistics. And if distinctions such
as that between langue and parole cannot be derived from the
first principles of an ambitious over-arching theory, they are
none the worse for that. The real test of their worth is the use
that they can be put to in the articulation of a theory of language
that makes the development of a coherent, criticisable meth-
odology for its study possible; and Saussure's distinctions cer-
tainly did that.
 NOTES

i. Saussure's work: its context and significance

The first page reference is to the second edition ofCours de linguistique
generate, the pagination of which is retained in other editions, in-
cluding de Mauro's critical edition (TM); the second reference is to
Wade Baskin's translation. Though the page numbering of Harris's
translation is different from either, the corresponding page of CLG
is always indicated there, so that it is possible to identify the relevant
page of his translation.
My discussion is particularly indebted to that of Chapter 1 of Ducrot
1968, which very clearly illustrates the value of placing Saussure in
a historical context. See also Culler 1976, ch. 3. A sympathetic ac-
count of the biological analogies developed by nineteenth-century
linguists can be found in Sampson 1980, ch. 1.
For a discussion of this see Chomsky 1966; Ducrot 1968, ch. 1;
Ducrot and Todorov 1972; Culler 1976, ch. 3.
The argument, in brief, is that since a given state of a language is
the unintentional by-product of individual speech acts, and since its
signs depend on their place within it, the latter are arbitrary in the
sense that they are historically constituted without being designed
or planned (5.2).
Although, as I said, it is not easy to see how the mixture could have
been avoided altogether, there is at least one major respect in which
the editors' way of combining the two terminologies is unfortunate.
In the first explanation given of a sign they retain Saussure's dis-
carded explanation of a sign as the union of two things, an acoustic

161
162 NOTES TO P P . 1 5 - 1 6
image and a concept, even though the latter term has individualistic
psychological connotations which are very misleading from the point
of view of the theory that Saussure goes on to develop, and quite
apart from the fact that the term has no intrinsic connections with
the term 'system'. It is true that later the new terms 'signifier' and
'signified' are introduced by the editors to replace the pair originally
used.
Ambiguity would disappear if the three notions involved here
were designated by three names, each suggesting and opposing
the others. I propose to retain the word sign [signe], to designate
the whole and to replace concept and sound-image respectively
by signified [signifie] and signifier [signifiant]: the last two terms
have the advantage of indicating the opposition that separated
them from each other and from the whole of which they are
parts. (CLG, 99, 67)
This passage suggests that, apart from the advantage of an implied
relationship, the two new terms have close family relationships with
the pair they replace, so that a signified is a concept. But in fact the
new terminology has radically different theoretical implications from
the old (6.1).
6. Examination of the sources reveals that this is yet another place at
which the initiative of the editors has been considerable. Apart from
the fact that the italicisation is their responsibility, for some reason
they discard the metaphor of the study of langue as a platform from
which one can view the position of other aspects of language (Ian-
gage). The metaphor is found in the notes of all the students. For
instance, Madame Sechehaye writes: '[It is] when one accords first
place to la langue, making it the point of departure, that one can
give their true place to the other elements of language (langage)'
(Engler 1, 30B). This certainly makes the study of la langue central,
in that it is the foundation of any kind of study of language, but
there is no suggestion that it is its own raison d'etre 'in and for itself.
So the italicised passage read in relation to its sources can hardly be
used to underwrite the interpretive summary of the whole of CLG
encapsulated in the final remarks.
7. As de Mauro points out, this order of presentation, which stresses
the interdependence of the study of languages and of language itself,
corresponded to a deep conviction of Saussure's which he articulated
in his 1891 inaugural lecture at Geneva:
... the most elementary linguistic phenomena will not be sur-
mised or clearly perceived, classified, and understood unless
one resorts in the first and last instance to the study of lan-
guages. ... On the other hand, to want to study languages ig-
noring the fact that they are governed primordially by certain
principles which are summarised in the idea of language (Ian-
gage) is a task even more bereft of any serious significance, of
any genuine scientific foundation. (TM, 354)
8. Here is a detailed comparison of the relevant passages:
 NOTES TO P. 16 163

Third Course CLG

(A)
Pt II: Language {la langue) Introduction:
Ch. I: La langue distinguished Ch. 3: The object of linguistics;
from langage ch. 4: Linguistics of language
and linguistics of speaking
(B)
Ch. II: Nature of the linguistic Pt 1: General principles
sign Ch. 1: Nature of the linguistic
sign
Ch. II': Immutability and Ch. 2: The same
mutability of the sign
Ch. II": Static and historical Ch. 3: The same
linguistics Pt 2: Synchronic linguistics
Ch. 1: Generalities [= (H)(i)]
(E)
Ch. Ill: The concrete entities Ch. 2: The same
of language Ch. 3: Identities, realities,
values [in part based on (E)]
(F)
Ch. 4: The abstract entities of Ch. 4: Linguistic value [in part
language based on (H)(iii)]
(G)
Ch. V: Absolute and relative Ch. 5: Associative and
arbitrariness syntagmatic relations [in part
based on (H)(ii)]
(H)
Ch. VI: Static linguistics Ch. 6: Mechanisms of language
(i) Generalities (SM III, 140-1) [6.1 is in part based on (H)
(ii); 6.3 on (G)]
(ii) Syntagmatic and associative Ch. 7: Grammar and its
relations (SM III, 142-7) subdivisions [based entirely
on the Second Course]
(iii) Values of terms and senses Ch. 8: Role of abstract entities
of words (SM III, i48ff) in grammar [= F]
9 Not everyone would agree about the importance of these issues.
Harris, for instance, writes, 'The question then is - and has been for
many years - how to make sense of reading this Saussure who is the
presumptive author of the Cours; not whether what we read is a
correct or incorrect account of "what the real Saussure really meant".
For whatever that may have been is arguably irrecoverable anyway'
(1987, viii). He goes on to argue that even if the Saussure of CLG
were a literary fabrication of the editors, he would be none the worse
for that; after all, 'so might Socrates, conceivably, be a fabrication
of Plato's' (ibid.).
164 NOTES TO PP. 19-21
2. The distinction between langue and parole
1. This is not an uncontroversial position. Culler, for instance, takes
a different view; see 1976, 34.
2. Often to talk of a discipline's object is to talk about the entities whith
which it is exclusively concerned, together with their essential prop-
erties. However, as de Mauro argues, the term 'object' does not
always have a material interpretation, as in 'What objects are being
talked about?' {TM> 414). Sometimes to talk about the object of a
discipline is to talk about what it seeks to explain. On this view, to
argue that langue is the object of linguistics is to argue that when
linguistic explanations are given, pride of place must be given to
the reconstruction of the langue (= underlying set of conventions)
of a language; but it does not follow that linguistic explanations
are concerned only with facts about langue. To explain how a par-
ticular langue is constituted, how it changes, and how it is acquired,
recourse may be necessary to facts of a quite different kind - ones
of parole (2.1.3). This is the interpretation of 'object' needed for a
proper understanding of Saussure's thesis that langue is the object
of linguistics.
3. There is a fourth term waiting to be fitted in also, viz. the individual
language faculty - faculte du langage.
4. This is not because I believe that a good command of French will
enable one to see easily what Saussure's distinctions really are; on
the contrary, he often uses ordinary words in a technical way. The
point is to ensure that it is always clear which of the three key terms
he has in mind.
5. For a discussion see Lyons 1968, 99ff. The first description would
be the subject matter of phonetics; the second, of phonology; so
speech sounds are the subject matter of the first discipline, whereas
it is the latter which is concerned with the functionally significant
features of the sounds. For instance, Lyons points out that in En-
glish p> t, and k in certain positions are slightly aspirated; but that
while this is of interest to phonetics, it is of no interest to phonology,
since 'the distinction between the aspirated and the unaspirated
variety never has the function of keeping apart different words in
English (this is a very crude, and partly inaccurate, statement that
will be treated more fully later); it is not a. functional difference: it
is a phonetic difference, but not a phonological, or phonemic dif-
ference of English' (p. 100).
I follow the convention which represents a phonetic description
of a speech sound by using square brackets, e.g. [nu]; a phonological
description is represented thus: /nu/.
6. For discussion of these issues and of the inadequacies of articulatory
and acoustic phonetics see Jakobson 1978, Lecture 1. Reporting
the work of Menzerath and Lacerda, Jakobson comments on its
implications for articulatory phonetics:
As for the speech chain, they arrived at an even more par-
adoxical conclusion. From a strictly articulatory point of view
 NOTES TO P P . 2 3 - 7 165
there is no succession of sounds However interesting or
important the study of linguistic sounds in their purely motor
aspect may be everything indicates to us that such a study is
no more than an auxiliary tool for linguistics, and that we
must look elsewhere for the principles by which the phonic
matter of language are organised, (p. 11)
He goes on to maintain that acoustic phonetics fares no better than
articulatory phonetics in providing principles of individuation:
Acoustics can provide us, in impressive detail, with the mi-
crographic image of each sound, but it cannot interpret this
image; it is not in a position to make use of its own results.
It is as if they were the hieroglyphics of an unknown language.
When, as is always the case, two sounds show both similarities
and dissimilarities, acoustics, having no intrinsic criteria for
distinguishing what is significant from what is not, has no
way of knowing whether it is the similarity or the dissimilarity
which is crucial in any given case. It cannot tell whether it is
a case of two variants of one sound or of two different sounds.
(P- 19)
Jakobson goes on to argue that this shows that 'we must ask what
is the immediate aim of sounds, considered as acoustic phenomena?
In raising the question we straight away go beyond the level of the
signifier, beyond the domain of sound as such, and we enter the
domain of the signified, the domain of meaning' (ibid.).
7. For a different interpretation see Harris 1987, 39.
8. Whether Saussure should have used this argument is another mat-
ter. It seems to imply that speech and writing are, or at least could
be, simply different ways of articulating the same language. But
this conflicts with the view that writing is a parasitic system and not
even part of the province of linguistics.
9. Some of the things that might have been included and that Saussure
lists are: 'the pure acoustical sensation, the identification of that
sensation with the latent sound-image, the muscular image of phon-
ation, etc' (CLG, 28, 12). Interestingly, there is no reference to
whatever thoughts either person might be having. Suppose, for
instance, A wanted to ask B for the time. Why should A opt for a
particular way of formulating the request, e.g., 'Have you got the
time?', rather than for one of the many other possible ways: 'Can
you tell me the time?', 'What time is it?', 'I'd like to know the time',
etc.? The model envisages the encoding of thoughts but gives no
account of the way in which they are formulated.
10. The text is far from clear, and it is possible that he did not mean
to count (i) as active and (v) as passive. In both cases all that is
involved is an 'unlocking', so that distinguishing them as respec-
tively active and passive seems arbitrary. On the other hand, he
clearly does want to say that reception is passive, for this is central
to his argument.
11. It is true that we have to wait for an account of this until later, in
fact until Part 2, Chapter 5; but the role staked out for it is clear
166 NOTES TO P P . 3 1 - 4 9
in outline. It is that faculty which enables B not merely to associate
with a given sound image the same concept as does A, but also to
relate the linguistic units identified in the same structurally signif-
icant ways, so that for instance B understands 'The dog bit the boy'
to say that it was the dog that bit the boy, and not vice versa. For
detailed discussion see 5.3.
12. The concept of generation is used here in the technical sense of
modern linguistics in which a grammar is said to generate a sen-
tence, and does not, therefore, imply anything about the corre-
sponding performance. In fact Saussure does not always count
sentences as units of parole. This is an issue which reveals consid-
erable tension in his thought; but I argue later that there is no
doubt that he was committed to treating sentences as part of langue
and that he had worked out a way of doing this in principle, though
certainly not in detail (5.3.2). However, since this is controversial,
for the time being I shall treat his claim that the sentence is not a
unit of langue as his considered view.
13. What Saussure meant by 'phonology' was not concerned with the
functionally significant features of sounds but with their articula-
tion, i.e., with what falls now within the province of phonetics. That
is why he treated phonology in his sense of the term as an auxiliary
science. By 'phonetics' he meant the study of the evolution of speech
sounds.
14. It should be said that despite apparently conceding that an ortho-
graphic representation can be as direct as an acoustic one, Lyons
argues for the priority of phonic substance (1968, 65).
15. If there were such a system, then the theoretical slack between the
written and the spoken forms would be completely taken up, and
we would no longer have to worry about how accurate a guide the
former was to the latter.

3. Language as a system of signs, I: Signs, arbitrariness,

linearity, and change
As Harris points out, a nomenclaturist might maintain (i) that there
are language-independent objects without maintaining that there
are language-independent ideas; (ii) that signs are psychological;
and (iii) that the relation between a name and what it names is
indirect, in that it is mediated by a sense (1987, 57).
Indeed, it is arguable that he went further and maintained that
there are no language-independent concepts, so turning the po-
sition of the Port Royal Grammar on its head (Ducrot 1968, 46).
However, since the rejection of the comparativists' claim does not
require Such a strong claim as this, we shall ignore it here.
His psychologistic stance is prima facie highly questionable ven if
a sign's intralinguistic relations are meaning-determining and it is
knowledge of them which enables members of the linguistic com-
munity to communicate with each other, it does not follow that the
relations in question have to be psychological ones holding between
 NOTES TO P P . 5 1 - 3 167
psychological entities. If the relations are meaning-determining, all
that is necessary for members of the linguistic community to com-
municate is that they should have similar representations of them,
not that the relations in question be ones relating psychological
entities.
The extent to which the new terminology has radical theoretical
implications which the original does not is clear from a discussion
later in CLG. Writing of signs as the union of a signifier and a
signified, Saussure says:
We constantly risk grasping only a part of the entity and
thinking we are embracing it in its totality; this would happen,
for example, if we divided the spoken chains into syllables,
for the syllable has no value except in phonology [ = modern
phonetics]. A succession of sounds is linguistic only if it sup-
ports an idea. Considered independently, it is material for a
physiological study, and nothing more than that.
The same is true of the signified ac soon as it is separated
from its signifier. Considered independently, concepts like
'house', 'white', 'see', etc. belong to psychology. They become
linguistic entities only when associated with sound-images; in
language a concept is a quality of its phonic substance just as
a particular slice of sound is a quality of the concept. (CLG,
144, 103)
But it is worth noting that it is one thing to maintain that there are
no language-independent meanings, and another to maintain that
signifiers belonging to different languages cannot have the same
meaning. The second claim follows from the first only if it is im-
possible for the signifiers to have the same meaning-determining
relations in their respective languages.
If I understand him, Benveniste has a different objection, namely
that to make his point Saussure has to introduce a third term,
'reality', as well as the two in question, the signifier and its signified;
but this is illegitimate if language is a form and not a substance
(1966, 50). Though Benveniste is right to say that Saussure's con-
cept of motivation takes into account the relation between some-
thing non-linguistic, e.g., a cuckoo, and its signifier and signified,
the concept is used to make a point central to his overall theory,
namely that there is nothing extralinguistic which determines the
nature of signifiers and signifieds. It is hard to see how such a point
could be made without doing what Benveniste says is illegitimate,
namely discuss the relations among signifiers, signifieds, and reality.
The point is well made in Burge 1975 that we must not so define
the conventional that for a convention C to exist there must also
exist an alternative C, which those who comply with C could adopt
here and now, as things are. What is required instead, as Burge
points out, is that it be true that there is an alternative C which,
if established, would have served the same purposes and which the
persons in question could have learned to conform to at an ap-
propriate stage of their development. For instance, once I have
168 NOTES TO P P . 5 4 - 8 0
learned to speak English with a Yorkshire accent I may find it
impossible to speak it with any other. But that does not mean that
my accent does not rest on usage and convention. Presumably a
different usage would have served broadly the same purposes, and
had I been exposed to it when a child I would have had a different
accent.
7. Someone who would quarrel with this conclusion is Jakobson. See
1978, 113.
8. Similarly, one could, as presumably Chomsky would, agree about
the marginality of onomatopoeia and interjections, and argue that
there is something external to a particular language, namely the
faculty of language, which motivates various features of it.
9. It might be argued that the numerical example is poorly chosen,
since the structure of the number descriptions in the language is
determined by the structure of the system of cardinal numbers. It
seems to me that this underestimates the idiosyncrasies of natural
language number systems. For instance, dix-sept is structure-
revealing, but seize is not, and quatre-vingts has a structure which is
not linguistically productive. Anyway, there are non-numerical ex-
amples, e.g., poirier, cerisier, pommier, etc. (CLG, 181, 131).
10. A good review and evaluation of these discussions is to be found
in Harris 1987, 69ff.
11. The relevant passage is CLG, 146, 104; it is discussed in 5.1.
12. I owe this point to John Rae.

4. Language as a system of signs, II: Diachronic and

synchronic linguistics
1. It is worth observing that at least one of the metaphors that Saussure
uses to illuminate the notion of a value, the economic metaphor,
does not sit very comfortably with the idea of linguistic values as
pure values (6.2).
2. Saussure says that 'changes are always unintentional, whilst the syn-
chronic fact is always significant' (CLG, 122, 85). This runs together
two different contrasts. On the one hand, he wants to say that syn-
chronic facts are always concerned with values, whereas diachronic
ones are not. But he also wants to deny that diachronic changes are
intended to produce the overall systematic effects they have. How-
ever, the rather curious remark about life being breathed into dif-
ferences suggests that he may have thought that values ultimately
do depend on intentions, e.g., on the collective will to find some
means of expressing the plural.
3. The other two analogies are the projection of an object onto a plane
surface and the difference between the structure revealed when a
plant is cut first transversely and then longitudinally (CLG, 125, 87).
4. In the case of chess, a state of the game may be defined as an
arrangement of pieces produced by a legitimate series of moves from
a given initial state. Clearly, no analogous explanation of what con-
stitutes a state of a given language is available.
 NOTES TO P P . 8 0 - 9 169
5. Lepschy makes a similar point: ' . . . there is a sense in which some
"diachronic" information is included in the rules of chess: one might
need to know whether the king or rook has moved, if one wants to
castle; one may need to know whether a pawn has just been moved
if one wants to take it en passant; and to decide whether a game ends
in a draw one needs to know whether the same position occurs for
the third time; or in the end game one may need to know how many
moves have been made from a certain moment onwards' (1970, 46).
6. Moreover, this response ignores the fact that in many states com-
petent players would make the same move, since what is a good
move is determined by the nature and purpose of the game and the
state in question. So in a given state of a game of chess played to
win, certain moves are much more likely than others. But then why
in a particular language state should there not be developments
which are more likely than others? For instance, in the example
discussed in 4.2, the fortuitous existence of two different ways of
marking the plural at Stage 2 in the case of the nouns in question
might make it more likely that sooner or later one or the other of
these would be used to mark the plural rather than some completely
different form. Lepschy makes a similar point: 'As traditional lin-
guistics used to repeat, language is in fact continuously changing,
even though the speaker may not be conscious of it and may translate
such changes as he experiences in terms of stylistic choices between
synchronically coexistent uses' (1970, 45).
7. That signs are concrete rather than abstract objects is a fundamental
part of Saussure's theory. This is so because they 'are realities that
have their seat in the brain' (CLG, 32,15). Moreover, they are tangible
in the rather peculiar sense that they can be 'reduced to convention-
al written symbols', so that if a language is a storehouse of sound
images, writing is its tangible form (ibid.). For further discussion
see 5.1.
8. Full discussion of this issue would take us too far afield. But, as
Harris points out in an insightful discussion of Saussure's theoretical
difficulties, he cannot allow individuals too much freedom over an-
alogical creation, because that would leave little or no explanatory
role for langue. On the other hand, he cannot overcircumscribe their
freedom, because in that case their only role would be to implement
changes implicit in a given state: 'Why are these extremes uncon-
genial from a Saussurean point of view? Because in the one case
langue merges with parole, and in the other synchrony merges with
diachrony' (Harris 1987, 152).

5. Language as a system of signs, III: Identities, system,

and relations
1. The principle would presumably rule that one cannot treat 'is
ing', which has the present continuous tense as its signified, as
an element of 'is running'. For an interesting discussion see Harris
1987, 111.
170 NOTES TO PP. 9 1 - 1 0 1
2. It is possible that Saussure's position about the psychological reality
of abstract entities was not fully worked out. For instance, in the
Third Course he expresses puzzlement about the point of the dis-
tinction between concrete and abstract entities. To consider an idea
apart from its signifier is to consider it as an abstraction; but, he
goes on to say, 'nothing in langue can be abstract if one takes the
concrete to be everything present to the consciousness of the speak-
ing subject' (SM, 84).
The possibility that a grammarian's analysis might involve dis-
tinctions not made by the native speaker is discussed in the appen-
dix to Parts 3 and 4 of CLG. This discussion contrasts two sorts of
analysis, the first called objective, which is made by grammarians,
and the second called subjective, which involves only distinctions
recognisable by the speaker. Saussure acknowledges that the two
sorts of analysis might diverge, but argues that they are equally
legitimate. This is certainly a puzzling remark, since it seems to
imply that an objective study is not necessarily one of langue. How-
ever, it may be that the views in question were ones he abandoned;
the material for the appendix comes from the First Course.
3. For an extended discussion of this and other sorts of difficulty see
Ducrot 1968, 48, and Harris 1987, io8ff.
4. The material discussed here is largely drawn from Part 2, Chapter
3, of CLG.
5. The material drawn on in this section comes almost entirely from
Chapters 5 and 6 of Part 2 of CLG. In these chapters the initiatives
of the editors were considerable. For instance, the expression 'the
set of phonic and conceptual differences' occurring in the first
sentence of Chapter 6 has no warrant in the text (Engler 2, 290).
The main effect of the editorial initiatives is the compression of
material which is much clearer in the notes than it is in their text.
6. The term 'discourse' is the editors'; in their notes the students all
use parole.
7. The examples are mine, but they are based on the ones Saussure
uses; cf. Engler 2, 280.
8. This was not a slip, for later we read: 'This notion of a syntagm
can be applied to unities of whatever kind; as much as to simple
words as complex ones and sentences' (Engler 2, 283).
9. Arguing that these relations are interdependent, Lyons writes that
the 'defining characteristic of modern "structural" linguistics is as
follows: linguistic units have no validity independently of their par-
adigmatic and syntagmatic relations with other units' (1968, 75).
Lyons uses the term 'paradigmatic' instead of Saussure's 'associa-
tive'. It should be said, though, that Lyons denies that syntagmatic
relations are necessarily sequential or time-ordered.
If there is an answer to the argument of 3.3 that it cannot be
true that the whole mechanism of language depends on the prin-
ciple of Linearity — even on Saussure's own account, since associative
relations are not linear - then it would, as far as I can see, have to
appeal to the interdependence of the two types of relation. But
 NOTES TO P P . 1 0 3 - 1 7 171
even so the principle would seem to be false if Lyons is right that
syntagmatic relations are not necessarily sequential.
10. This consists of a remark to the effect that we should recognise in
contrelmarche a relation between its parts (contre and marche), and
also a part-whole relation, viz. that between contre, marche, and
contremarche. In the first case we are concerned with a relation of
succession; and in the second, one which is productive in the sense
that contremarche has to be seen as the product of its parts (Engler
2, 283).

6. Language as a system of signs, IV: Values, differences,

and reality
1. The order in which I discuss these is that of the Third Course,
which is the reverse of CLG.
2. If Saussure is right, mouton doubles as a count noun and a bulk
term. For some reason, 'sheep' does not, though 'lamb' does -
compare 'Mary had a little lamb' with the continuations 'to eat' and
'which followed her to school one day'. Qua count noun, should
we say that 'lamb' has the same signified as does 'lamb' qua bulk
term? There are good Saussurean reasons for answering in the
negative; the count noun is potentially modifiable in ways in which
the bulk term is not, and vice-versa. Moreover, there are equally
cogent reasons for distinguishing the signifieds of mouton in its two
uses. So it is not clear why 'sheep' and mouton should not have the
same signified provided the latter is not being used as a bulk term.
3. The diagram is a modification of one that appears in the students'
notes:

In this diagram langue is intermediate between thought and sound

but is not a modification of either. The divisions introduced by
langue are clearly not phonic on this model; it is not clear that they
are not in the one devised by the editors.
4. The passage is in fact a much-compressed version of a longer one
from the Second Course which the editors inserted into the Third
Course at this point. One can see why they felt the need to insert
some account of the origin of systems of values at this stage.
5. For a general discussion of these issues and of the theory of semantic
fields see Lehrer 1974 and Lyons 1977, ch. 8.
6. For instance, I argued earlier that one of the defects of the way in
which Saussure attempts to define the langue/parole distinction is
that he assumes that there is a given field langage whose boundaries
172 NOTES TO P P . 1 1 7 - 2 6
are clear, so that it can be divided. On this account, parole is what
is left when langue has been subtracted. But since we do not know
precisely what belongs to langage, we do not know what the re-
mainder is (2.1.3).
7. For a critical discussion of the view that underlying all vocabularies
there is an unstructured thought mass or meaning substance see
Lyons 1977, 259.
8. For a discussion and evaluation of it see Lyons 1977, 246.
9. Here 'phoneme' is being used in the modern sense for a sound
which is linguistically functional. For a discussion see TM, nn. 236,
111. See also Lyons 1968, 9gff.
10. The clause begining 'it is even' is an addition of the editors'. The
passage reveals considerable tension in Saussure's thought. The
sign is said to be the union of a signifier and signified, and is as
such a concrete entity. From this point of view, to consider the
signifier or signified in isolation is to consider it as an abstraction.
On the other hand, Saussure argues that a langue is an association
of two independent orders of differences which, as Caws says,
'stresses the incommensurability, perhaps the mutual inaccessibility
of the two domains, which have no essential connection but only
an arbitrary and artificially forced association' (1988, 75).
11. It is important to note that the idea of a lexical field, which is
important for linguistics, does not presuppose that anything is
given. Consider, for instance, Lyons's explanation of a lexical
field:
Lexemes and other units that are semantically related,
whether paradigmatically [ = Saussure's 'associatively'] or syn-
tagmatically, within a given language system can be said to
belong to, or to be members of, the same (semantic) field;
and a field whose members are lexemes is a lexical field. A
lexical field is therefore a paradigmatically and syntagmati
cally structured subset of the vocabulary (or lexicon). (1977,
268)
Clearly, this account does not assume that the field is given. On the
contrary, what fields there are depends on what paradigmatic and
syntagmatic relations there are between lexemes.
12. This is not, however, always true; there is surely a systematic dif-
ference between 'butcher' and 'baker' within the domain of occu-
pations, and between 'bite' and 'scratch' in the domain of aggressive
acts, yet 'X is a butcher' does not imply 'X is not a baker' any more
than 'X bit Mary' implies 'X did not scratch Mary'. Great care is
needed in stating a rule that covers cases like these or, indeed,
applies to other tenses and verbs; for instance, 'X will be hot' does
not imply 'X will not be cold' any more than 'X went on Saturday'
implies 'X did not go on Tuesday' (Lehrer 1974, 25).
13. These are not meant to be exhaustive. For a more detailed discus-
sion see Lyons 1977, ch. 9, and Lehrer 1974, ch. 2. Of particular
interest is the type of opposition which Lyons calls 'direc-
tional'. This
 NOTES TO P P . 1 2 7 - 5 3 173
is seen most clearly in the relationship which holds between
'up': 'down', 'arrive': 'depart', and 'come': 'go'. What these
pairs have in common, in what might be regarded as their
most typical usage, is an implication of motion in one of two
opposed directions with respect to a given place P. (1977,
281)
The challenge from the point of view of structuralist semantics
would be to account for the apparently essential way in which
the use of these terms involves reference to a position.
14. For a discussion of the terms 'marked' and 'unmarked' see Lyons
1977, 305. The binarist would of course have to explain how to
reduce the many apparently non-binary contrasts there are to
sets of binary contrasts.
7. Successes and failures
1. Of course the tradition which has influenced Quine's views is not
structuralism but pragmatism, and in particular the work of Dewey.
Even so, Dewey reached a number of conclusions which would have
been congenial to Saussure. For Dewey, 'language is specifically a
mode of interaction of at least two beings, a speaker and a hearer;
it presupposes an organized group from which they have acquired
their habits of speech. It is, therefore, a relationship' (1958, 185).
On the other hand, Dewey and Quine's behaviourism would not
have been acceptable to Saussure.
2. For a detailed discussion of this see Gibson 1982 and Gochet 1986.
3. Roughly speaking, the difference between underdetermination and
indeterminacy is that whilst in the first case the data are compatible
with different hypotheses, nevertheless one of them is correct; in
the second this is not so. In other words, in the first case there is
a fact of the matter; in the second there is none.
4. For Quine's conception of an observation sentence see Gochet 1986,
32-4,44-6.
5. For valuable discussions of the issues involved see O'Neill 1973 and
MacDonald & Pettit 1981.
6. The latter suggestion has some historical provenance. Louis Caille,
for instance, claimed that the comparison Saussure makes of lin-
guistic value with forms of exchange was suggested by Tarde's
Psychologie economique; others have suggested that the distinction
between langue and parole is to be found in Tarde's Lois de Vimitation.
7. For a valuable and very fair discussion of this issue see D'Agostino
1986, 37ff.
8. A fuller, though still somewhat abbreviated, version of his analysis
is as follows:
A regularity R is a convention in a population P, if and only
if, within P
(1) Everyone conforms to R.
(2) Everyone believes that the others conform to R.
(3) The belief that the others conform to R gives everyone
a good and decisive reason to conform to R himself.
174 NOTES TO PP. 1 5 4 - 6
(4) There is a general preference for general conformity
to R rather than slighly-less-than-general-conformity - in par-
ticular conformity by all but any one.
(5) There is at least one alternative regularity R' which is
such that the beliefs that the others conformed to R' would
give everyone a decisive reason to conform to R'.
(6) (i)-(5) are matters of mutual knowledge: they are
known to everyone, it is known to everyone that they are
known to everyone, and so on. (Lewis 1975, 5)
9. The role of the mutual-knowledge conditions in Lewis's account is
not uncontroversial. Burge has argued convincingly that the exis •
tence of an alternative need not be mutually known:
Imagine a small, isolated, unenterprising linguistic commu-
nity none of whose members ever heard anyone's speaking
differently. Such a community would not know - or perhaps
have reason to believe - that there are humanly possible al-
ternatives to speaking their language Yet we have no in-
clination to deny that their language is conventional. (Burge
i975» 250)
Furthermore, in Lewis's detailed analysis (n. 8 above), provided
conditions (2) and (3) are satisfied, everyone has a reason to con-
form whether or not (6) is satisfied. But perhaps without (6) that
reason would be a weak one.
10. For a discussion of semiology and linguistics see Barthes 1967;
Chomsky 1969, ch. 3; Robey 1973; Culler 1976, ch. 4; Leiber 1978;
Caws 1988.
11. Quite how this is to be done in detail is not clear, but Culler has
suggested the broad lines which such a derivation might take:
Saussure's notes suggest [that] the distinction between langue
and parole is a logical and necessary consequence of the ar-
bitrary nature of the sign and the problem of identity in
linguistics. In brief: if the sign is arbitrary, then, as we have
seen, it is a purely relational entity, and if we wish to define
and identify signs we must look to the system of relations and
distinctions which create them. We must therefore distinguish
between the various substances in which signs are manifested
and the actual forms which constitute signs; and when we
have done this what we have isolated is a system of forms
which underlies actual linguistic behavior or manifestation.
The system of forms is la langue; the attempt to study signs
leads us, inexorably, to take this as the proper object of lin-
guistic investigation. (1976, 34)
However, for reasons given later in this chapter, this derivation
would seem to be somewhat shaky.
12. These examples are only suggestions. But an examination of the
design features of language does not encourage either the idea that
it is just one sign system amongst many or the idea that study of
other sign systems will throw a great deal of light on it (Lyons 1977).
 BIBLIOGRAPHY

Sources and texts

Saussure, Ferdinand de. 1916. Cours de linguistique generate, edited by
Charles Bally and Albert Sechehaye, with the collaboration of Al-
bert Riedlinger. Lausanne and Paris. (Page references are to the
second edition of 1922, cited as CLG. This pagination is retained
by subsequent editions, including de Mauro's critical edition.)
1972. Cours de linguistique generate. Critical edition, edited by Tullio
de Mauro. Paris: Payot. (Cited as TM)
1977. Course in General Linguistics, translated by Wade Baskin. Glas-
gow: Fontana/Collins. (Cited as WB)
1986. Course in General Linguistics, translated by Roy Harris. La Salle,
111.: Open Court.
Engler, Rudolf. 1967. Cours de linguistique generate: edition critique. 3
fascicules, continuously paginated. Wiesbaden: Otto Harrassowitz.
(Cited as Engler. The left-hand column gives the text of CLG, and
the other five its sources, which are referred to by use of 'B' to T \
Thus 'Engler 2, 148 F' refers to the last column of p. 148 in the
second fascicule.)
Godel, Robert. 1954. 'Notes inedites de F. de Saussure'. Cahiers F. de
Saussure 12, 49—71.
1957. Les sources manuscrites du cours de linguistique generate. Geneva:
Librairie Droz. (Cited as SM)
Godel, Robert (ed.). 1969. A Geneva School Reader in Linguistics. Bloom-
ington: Indiana University Press.

Other works cited

Aarsleff, H. 1967. The Study of Language in England, 1760—1860. Prince-
ton, N.J.: Princeton University Press.

175
176 BIBLIOGRAPHY
Amacker, R. 1975. Linguistique saussurienne. Geneva and Paris: Librairie
Droz.
Barthes, R. 1967. Elements of Semiology, translated by A. Lavers and C.
Smith. New York: Hill and Wang.
Benveniste, E. 1966. Problemes de linguistique generate. Paris: Gallimard*
Berlin, B., and P. Kay. 1970. Basic Color Terms. Berkeley and Los An-
geles: University of California Press.
Burge, T. 1975. 'On Knowledge and Convention'. Philosophical Review
84, 249-55.
Calvet, L. 1975. Pour et contre Saussure. Paris: Payot.
Caws, P. 1988. Structuralism: The Art of the Intelligible. Atlantic Highlands,
N.J.: Humanities Press.
Chomsky, N. 1957. Syntactic Structures. The Hague: Mouton.
1964. Current Issues in Linguistic Theory. The Hague: Mouton.
1965. Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press.
1966. Cartesian Linguistics: A Chapter in the History of Rationalist Thought.
New York: Harper and Row.
1969. Language and Responsibility. Hassocks: Harvester.
Culler, J. 1976. Saussure. Glasgow: Fontana/Collins.
D'Agostino, F. 1986. Chomsky's System of Ideas. Oxford: Clarendon Press.
Danto, A. C. 1973. 'Methodological Individualism and Methodological
Socialism'. In O'Neill 1973, 312—37.
Derrida, J. 1976. OfGrammatology, translated by G. C. Spivak. Baltimore:
Johns Hopkins University Press.
1981. Positions, translated and annotated by A. Bass. London:
Athlone.
Dewey, J. 1958. Experience and Nature. La Salle, 111.: Open Court.
Doroszewski, W. 1933. 'Quelques remarques sur les rapports de la so-
ciologie et de la linguistique', Journal de Psychologie.
Ducrot, O. 1968. Le structuralisme en linguistique. Paris: Seuil.
Ducrot, O., and T. Todorov. 1972. Dictionnaire encyclopedique des sciences
du langage. Paris: Seuil.
Durkheim, E. 1982. The Rules of Sociological Method, edited with an
introduction by S. Lukes; translated by W. D. Halls. London:
Macmillan.
1952. Suicide: A Study in Sociology, translated by J. A. Spaulding and
G. Simpson. London: Macmillan.
Firth, J. R. 1957. Papers in Linguistics. Oxford: Oxford University Press.
Gibson, R. F. 1982. The Philosophy ofW. V. Quine. Tampa: University
of South Florida.
Giddens, A. 1978. Durkheim. Hassocks: Harvester.
Gochet, P. 1986. Ascent to Truth: A Critical Examination of Quine's Philos-
ophy. Munich: Philosophia Verlag.
Goffman, E. 1972. Relations in Public. Harmondsworth: Penguin.
Harris, R. 1987. Reading Saussure. London: Duckworth.
 BIBLIOGRAPHY 177
Henry, A. 1970. 'La linearite du signifiant'. In J. Dierickx and Y. Lebrun
(eds.), Linguistique contemporaine: hommage a Eric Buyssens. Brussels:
Editions de L'Institut de Sociologie, 86—92.
Jakobson, R. 1978. Six Lectures on Sound and Meaning, translated by J.
Mepham. Hassocks: Harvester.
Lehrer, A. 1974. Semantic Fields and Lexical Structure. Amsterdam: North-
Holland.
Leiber, J. 1978. Structuralism: Skepticism and Mind in the Psychological
Sciences. Boston: Twayne.
Lepschy, G. C. 1970. A Survey of Structural Linguistics. London: Faber
and Faber.
Levi-Strauss, C. 1977. Structural Anthropology, translated by C. Jacobson
and B. Grundfest Schoepf. Harmondsworth: Penguin.
Lewis, D. 1975. 'Language and Languages'. In K. Gunderson (ed.),
Language, Mind and Knowledge. Minnesota Studies in the Philosophy
of Science, VII. Minneapolis: University of Minnesota Press, 3-35.
Lukes, S. 1973. Durkheim. London: Allen Lane.
Lyons, J. 1968. An Introduction to Theoretical Linguistics. Cambridge: Cam-
bridge University Press.
1977. Semantics, vol. 1. Cambridge: Cambridge University Press.
MacDonald, G., and P. Pettit. 1981. Semantics and Social Science. London:
Routledge and Kegan Paul.
Mill, J. S. 1875. A System of Logic, 9th ed. London: Longman.
Mounin, G. 1968. Saussure ou le structuralismesans lesavoir. Paris: Editions
Seghers.
O'Neill, J. (ed.). 1973. Modes of Individualism and Collectivism. London:
Heinemann.
Robey, D. (ed.). 1973. Structuralism: An Introduction. Oxford: Clarendon
Press.
Quine, W. V. i960. Word and Object. New York: Wiley.
1969. Ontological Relativity and Other Essays. New York: Columbia Uni-
versity Press.
1977. 'Facts of the Matter'. In R. S. Shahan and C. Swoyer (eds.),
Essays in the Philosophy of Quine. Hassocks: Harvester, 155-69.
Sampson, G. 1980. Schools of Linguistics: Competition and Evolution. Lon-
don: Hutchinson.
Schleicher, F. 1869. Darwinism Tested by the Science of Language, translated
by A. Bikkers. London: Hotten.
Trier, J. 1931. Der deutsche Wortschatz im Sinnbezirk des Verstandes. Hei-
delberg: Winter.
1934. 'Das sprachliche Feld', Jahrbuch fur Deutsche Wissenschaft 10.
Watkins, J. 1973. 'Ideal Types and Historical Explanation'. In O'Neill
 Index

abstract entities of language, 83, 89— comparative linguistics, 6-9, 17, 4 1 -

92, 170 2, 48-9, 70-1, 136-7, 155
acoustic image, 14, 35, 50—1, 99—100 concepts, 14, 50-1, 108-11, 162
arbitrary nature of the sign, 12, 52- concrete entities of language, 83, 89-
5, 61, 64, 73, 94, 112-15, 120, 92, 134, 170
135, 138, 143, 157-9, l 6 l 5 a b s o " convention, 153—4, 167, 173
lute and relative arbitrariness, Culler, J., 44-5, 74, 124, 135, 137
55-6, 93-7; and motivation, 5 3 -
5> 167 D'Agostino, F., 173
associative relations, 57, 60, 83, 99- delimitation of linguistic units, 89-
104 92; indeterminacy of, 90, 105,
136, 140-3
Bally, C, 1, 13-16, 51, 136-7, 161-2, Derrida, J., 39—40
170 diachronic linguistics, 17, 69-71, 74-
Barthes, R., 5, 159 8, 81-5, 139
Benveniste, E., 52-3, 62, 167 differences and linguistic value, 95—
Berlin, B., 117, 132 7, 119, 121—30
Bopp, F., 7 Dorozewski, W., 144
Broca, P., 24 Ducrot, O., 136, 140, 143
Burge, T., 167, 174 Durkheim, E., 8; and Saussure,
144-7
Caille, L., 144, 173
Calvet, L., 14-16 editors of CLG, 1, 13-16, 51, 136-7,
Caws, P., 172 161-2, 170
Chomsky, N., 31, 94, 139, 168 Engler, R., 14
colour terms, 115-19
communication: Saussure's model, Godel, R., 14
24-6, 46, 49 Goffman, E., 157

178
 INDEX
Harris, R., 41-2, 48, 58-9, 163, 166,
169
historical linguistics, see comparative
linguistics
Hjelmslev, L., 44, 137
identity, synchronic, 22, 92, 95-7,
104-5, 1X9
interjections, 54
Jakobson, R., 52, 58, 60, 121-3, 132,
164-5
Kay, P., 117, 132
language, 4, 22-3, 134-5; autonomy
of, 10-13, 17, 55, 65; compari-
son with chess, 76—8, 114—15,
139, 168-9; comparison with
Morse Code, 34, 37, 44-5; com-
parison with a symphony, 33, 37,
44—5; faculty of, 23-4, 42-4,
168; and social facts, 8, 22, 27-
30, 61-3, 65, 145-55; a n d
thought, 112-15; see also abstract
and concrete entities of language
langue, 15, 21-4, 26-30, 34-6, 44-5,
65—6, 84—7, 151—2; and associa-
tive and syntagmatic relations,
100—4; a s a form, 36-9, 112-15;
and parole, 20—1, 33—4, 45—6,
Lehrer, A., 116, 128
Lepschy, G., 169
Levi-Strauss, C , 5, 53, 102, 158
Lewis, D., 153-4, 173-4
lexical fields, 116-17, 172
linearity of the signifier, 56—61, 89,
94, 98, 137, 159
linguistics, 15, 17, 73, 134-7, 160;
history of, 6-9; object of, 19-20,
22-3, 34-6, 46, 81, 84-5, 164;
see also comparative, diachronic,
structural, and synchronic
linguistics
Lukes, S., 146
Lyons, J., 38, 94, 125-7, 130, 133,
164, 170, 172
Mauro, T. de, 14-15, 162, 164
Meillet, A., 144
methodological individualism, 147—9;
and Saussurean linguistics, 149-
52, 154-5
'79
Mill, J. S., 147-8
Mounin, G., 5, 8, 144
Neo-Grammarians, 9
nomenclaturism, 11-12, 48-50, 66,
113, 142
onomatopoeia, 54
oppositions, linguistic, 110, 112, 121,
125-7; binary, 127-9
parole, 30-4, 84-5, 87, 138; and
langue, 20-1, 45-6, 137-9
philosophers, Saussure's advice to,
phonemes, 120-3
Popper, K., 151
Port Royal Grammar, 10—12, 17, 55,
142, 166
Quine, W. V., 140-3
Rae,J., 168
Riedlinger, A., 13
Schleicher, F., 7-9
Sechehaye, A., 1, 13-16, 51, 136-7,
161—2, 170
semiology, 5-6, 17, 19, 35-6, 59, 61,
79> 83> 94> 98> i37» 155~Qo
signified, 14, 50—5, 108—11, 119, 162;
and differences, 121-30
signifier, 14, 50-5, 56-61, 165; see
also linearity of the signifier
signs, linguistic, 49—52, 155—60, and
see arbitrary nature of the sign;
form a system, 93-7, 108-10,
119, 125, 135; immutability of,
62-3; mutability of, 63-6; see also
value: linguistic
social facts, 8, 27-8, 61-3, 135, 145-
7; and Saussurean Linguistics,
152-5
social realism, 143, 148; and Saus-
sure, 152-5
speech sounds, 21-2, 40-3, 89,
164-5
structural linguistics, 104, 130,
170
structuralism, 5, 130, 135
synchronic linguistics, 17, 69-71, 74-
8, 81-5, 134-5, 139; is really
idiosynchronic, 82; see also iden-
tity, synchronic
180 INDEX
syntagmatic relations, 57, 60, 83, 9 8 -
100; relations to langue and pa-
role, 100-4
Tarde, G., i44~7> ^ 3
thought-sound, 112-15,117,124
translation, 110; radical, 140-1
Trier, J., 116
value: linguistic, 51, 71-4, 77-9, 104,
108—9, 12i—4; and differences,
95~7> 119» x 21-30; and ex-
change, 108-12; and the indis-
tinctness of pre-linguistic
thought, 112-15; kinds of,
120-1; see also oppositions,
linguistic
Watkins,J., 148
Whitney, W. D., 8, 23, 144
writing, 39-42, 156